10.1.1.200.3302.pdf

ACTS Project 215Cellular Radio Access for Broadband Services

CRABS

D2P1B

SPECIFICATION OF NEXT-GENERATION OF LMDSARCHITECTURE

Project Number: AC215

Project Title: CRABS

Deliverable Type: K

CEC Deliverable Number: AC215/TEL/RD/DR/P/D2P1B/b1

Contractual Date of Delivery to the CEC: Nov. 1998

Actual Date of Delivery to the CEC: .2. February 1999

Title of Deliverable: Specification of next-generation of LMDS architecture

Workpackage contributing to the Deliverable: WG2

Nature of the Deliverable: R

Abstract:This report describes the specification of the next generation LMDS system. It contain a description ofthe overall system as well as detailed simulation and anlysis of radio sub-systems and protocols.

Keyword list: Broadband wireless access, LMDS radio sub-systems, millimetre radio propagaion,modulation and coding schemes, frequency plans and re-use strategies.

Editor: H. Loktu (TEL)

Contributions from: I. Bucaille (TOM), A. Burgess (EBL), G. Coppola (PRM), D. Evans (PRL), V.Ferrero (POL), R. Germon (PBN), O. Koudelka (IAS), T. Kourtis (DEM), A. Nordbotten (TEL), H.Loktu (TEL), J. Norbury (RAL), M. Orifice (POL)

AC215/TEL/RD/DR/P/D2P1B/b1 Deliverable Report D2P1B

i

SUMMARY

In Project 215 Cellular Radio Access to Broadband Services (CRABS) in the ACTS program,Working Group 2 has dealt with issues related to system architecture and simulation studies. The mainobjectives has been to study and specify the system architecture for the next generation system ofwireless broadband access systems with particular focus on Local Multipoint Distributions Systems(LMDS). Next generation systems within this context is understood as the commercial equipment andsystems available on the market within 1 to 2 years from finalisation of the project.

Targetting a combined residential and SME market scenario, a medium to high interactivity terminalwas identified as the major design objective. This was concluded to be very suitable for the high-end ofthe residential customers, small offices or home offices and the low-end of the SME customers.Furthemore, to enter the residential market a low-cost approach to terminal design is regarded asmandatory. To achieve this goal extensive re-use of widely supported standards assumed to a necessarylimitation in devising a solution within the time frame of the next generation of LMDS equipment.

The proposed next generation LMDS system is capable of supporting a full service access network withuser bit rate better than 8 Mb/s on the uplink and 34 Mb/s dowlink. Using the DVB as the baselinestandard familiy, low-cost user terminals will most likely occur. It is designed for quasi-error-freeoperation and offers availability figures better than 99.99 %. A cellular approach is recommended withtypically 2000 residential customers within a single cell.


ii

TABLE OF CONTENTS

SUMMARY................................................................ ................................ ................................ ..................... I

TABLE OF CONTENTS................................................................ ................................ .............................. II

GLOSSARY OF TERMS AND ACRONYMS................................................................ ............................. V

1. INTRODUCTION................................................................ ................................ ................................ ..... 1

2. SYSTEM REQUIREMENTS AND DESIGN OBJECTIVES ................................................................ . 2

2.1 EVOLUTION OF MVDS AND LMDS IN AN EUROPEAN PERSPECTIVE .............................................................. 2

2.2 TARGET LMDS USER CLASSES AND SEGMENTS ................................................................ ............................ 2 2.2.1 User classes................................................................ ................................ ................................ ....... 3 2.2.2 Geographic user segmentation................................................................ ................................ ........... 4

2.3 LMDS SERVICE DEFINITIONS ................................................................ ................................ ...................... 5 2.3.1 Classification of target services ................................................................ ................................ ......... 6 2.3.2 User class service configuration ................................................................ ................................ ........ 7 2.3.3 Definition of network availability................................................................ ................................ ....... 8

2.4 LMDS TERMINAL CONFIGURATIONS................................................................ ................................ ............ 9 2.4.1 IDU and ODU terminal requirements ................................................................ ................................ 9 2.4.2 User classes terminal options................................................................ ................................ ............. 9

3. LMDS SYSTEM ARCHITECTURE ................................................................ ................................ ..... 11

3.1 OVERVIEW OF LMDS SYSTEM ARCHITECTURE ................................................................ ........................... 11

3.2 INTERNAL TRANSPORT NETWORKS................................................................ ................................ ............. 12 3.2.1 Connection to external networks ................................................................ ................................ ...... 12 3.2.2 LMDS base station feeder network................................................................ ................................ ... 13 3.2.3 Microwave BS feeder networks ................................................................ ................................ ........ 15

3.3 CELLULAR LMDS ACCESS NETWORK ................................................................ ................................ ........ 18 3.3.1 The concept of a LMDS cell ................................................................ ................................ ............. 18 3.3.2 Cellular subscriber coverage ................................................................ ................................ ........... 19 3.3.3 Cellular network capacity ................................................................ ................................ ................ 20 3.3.4 Access network availability................................................................ ................................ .............. 23

3.4 IN-HOUSE DISTRIBUTION NETWORKS (IHDN)................................................................ ............................. 25 3.4.1 Wireline options and requirements for IHDN................................................................ ................... 25 3.4.3 Radio Local Area Network (RLAN) ................................................................ ................................ .. 30

4. FREQUENCY PLANS AND REUSE STRATEGIES ................................................................ ........... 33

4.1 INTRODUCTION ................................................................ ................................ ................................ ........ 33

4.2 FREQUENCY PLAN................................................................ ................................ ................................ ..... 33 4.2.1 Dual frequency/dual polarisation plan................................................................ ............................. 33 4.2.2 Channel plan for downlink................................................................ ................................ ............... 34 4.2.3 Two or more operators in the same franchise area................................................................ ........... 36

4.3 SERVICE MAPPING ONTO FREQUENCY PLANS................................................................ ............................... 36 4.3.1 Broadcast channels................................................................ ................................ .......................... 36 4.3.2 ATM data services ................................................................ ................................ ........................... 38

4.4 MULTI-LOBE ANTENNAS TO ACHIEVE GREATER CAPACITY IN EACH CELL................................. ..................... 41

4.5 FREQUENCY PLAN INTERFERENCE ISSUES ................................................................ ................................ ... 42 4.5.1 Interference between near adjacent cells ................................................................ ......................... 42 4.5.2 Frequency allocation schemes to limit interference in near adjacent cells ....................................... 43 4.5.3 The effect on interference situations when using diversity improvement ........................................... 44

4.6 OTHER FREQUENCY BAND ISSUES ................................................................ ................................ .............. 44 4.6.1 Bandwidth allocated to backbone system ................................................................ ......................... 44 4.6.2 Alternative cell plan which produces better C/I values................................................................ ..... 44 4.6.3 Band plan for the extra 1 GHz of spectrum from 42.5 to 43.5GHz.................................................... 46

5. LMDS RADIO SUB-SYSTEMS................................................................ ................................ .............. 49


iii

5.1 RADIO SUB-SYSTEM DESIGN FRAMEWORK ................................................................ ................................ .. 49

5.2 TRANSMISSION AND MULTIPLEXING................................................................ ................................ ........... 51 5.2.1 Multiple access schemes and framing structures ................................................................ .............. 51 5.2.2 Modulation and coding schemes ................................................................ ................................ ...... 52

5.3 ANTENNA CHARACTERISTICS ................................................................ ................................ .................... 55 5.3.1 Directive Base Station antennas................................................................. ................................ ...... 55 5.3.2 Omni-directional Base Station antennas ................................................................ .......................... 57 5.3.3 User Terminal antennas................................................................ ................................ ................... 58 5.3.4 Base station antennas with more complex pointing system. .............................................................. 59 5.3.5 Bandwidth considerations ................................................................ ................................ ................ 60 5.3.6 Dual linear (H/V) polarisation antennas ................................................................ .......................... 60

5.4 MILLIMETRE-WAVE EQUIPMENT AND TECHNOLOGIES AT 40 GHZ ................................................................ 60 5.4.1 Base station ................................................................ ................................ ................................ ..... 60 5.4.2 Consumer Premise Equipment (CPE)................................................................ ............................... 62 5.4.3 Review of 40 GHz MMIC Technologies................................................................ ............................ 65

6 RADIO MEDIUM ACCESS AND CONTROL ................................................................ ...................... 69

6.1 PROTOCOLS FOR MEDIUM ACCESS (MAC)................................................................ ................................ . 69 6.1.1 LISSY................................................................ ................................ ................................ ............... 69 6.1.2 C-TDMA................................................................ ................................ ................................ .......... 71 6.1.3 DAVIC................................................................ ................................ ................................ ............. 71

6.2 PROTOCOL PARAMETERS COMPARISON ................................................................ ................................ ...... 72

6.3 SIMULATION RESULTS COMPARISON................................................................ ................................ ........... 73 6.3.1 Simulation hypothesis ................................................................ ................................ ...................... 74 6.3.2 End-to-end delay comparison ................................................................ ................................ .......... 75

6.4 CONCLUDING REMARKS ................................................................ ................................ ............................ 76

7. CONLUSIONS AND RECOMMENDATIONS................................................................ ..................... 78

REFERENCES................................................................ ................................ ................................ ............. 80

APPENDIX A REGULATORY ISSUES, STANDARDS AND REFERENCE MODELS ....................... 82

A.1 INTRODUCTION................................................................ ................................ ................................ ........ 82

A.2 SPECTRUM ISSUES................................................................ ................................ ................................ .... 82 A.2.1 CEPT ................................................................ ................................ ................................ .............. 82 A.2.2 WRC................................................................ ................................ ................................ ................ 82 A.2.3 ETSI Project BRAN Spectrum Matters................................................................ ............................. 83 A.2.4 DECT................................................................ ................................ ................................ .............. 83

A.3 MPT1560 ................................................................ ................................ ................................ ............... 83

A.4 ETSI AND DVB STANDARDS ................................................................ ................................ ................... 83 A.4.1 DVB ................................................................ ................................ ................................ ................ 83 A.4.2 DVB LMDS Interaction Channel Specification ................................................................ ................ 84

A.4.3 Other Relevant DVB Specifications ................................................................ ................................ . 86 A.4.4 DECT................................................................ ................................ ................................ .............. 86 A.4.5 Additional ETSI Specifications ................................................................ ................................ ........ 87

A.5 DAVIC................................................................ ................................ ................................ ................... 87 A.5.1 Introduction ................................................................ ................................ ................................ .... 87 A.5.2 DAVIC LMDS Specification ................................................................ ................................ ............ 88

A.6 ETSI PROJECT BRAN ................................................................ ................................ ............................. 90 A.6.1 Introduction ................................................................ ................................ ................................ .... 90 A.6.2 HIPERACCESS ................................................................ ................................ ............................... 91

A.7 NETWORK ARCHITECTURE ................................................................ ................................ ....................... 91

A.8 REFERENCE MODELS ................................................................ ................................ ............................... 92

A.9 SUMMARY ................................................................ ................................ ................................ ............... 92

APPENDIX B - SUMMARY OF SIMULATION RESULTS................................................................ ..... 93

B.1 INTRODUCTION ................................................................ ................................ ................................ ........ 93

B.2 DOWNLINK RECEIVER CONSIDERATIONS ................................................................ ................................ ... 93

B.3 ANTENNAS ................................................................ ................................ ................................ .............. 94


iv

B.4 ARCHITECTURE FOR MASTER ANTENNA PASSIVE OPTICAL NETWORKS....................................................... 94

B.5 INTERACTIVE UPLINK................................................................ ................................ ............................... 94

B.6 CHANNEL ................................................................ ................................ ................................ ................ 94

B.7 NETWORK CONNECTION ................................................................ ................................ .......................... 95


v

GLOSSARY OF TERMS AND ACRONYMS

AAL ATM Adaptation LayerABR Available BitRateACTS Advanced Communication Technologies and ServicesATM Asynchronous Transfer ModeAWGN Additive White Gaussian NoiseBAP Broadband Access PointBER Bit Error RatioBISDN Broadband Integrated Services Digital NetworkBoD Bandwidth on DemandBS Base StationBWA Broadband Wireless AccessC4 Communication, Computing, Contents and Consumer electronicsCBR Constant BitRateCDMA Code Division Multiple AccessCEPT Conferenc Europenne des Postes et Tlgraphes et TlcommunicationsCORP CorporationCNR Carrier-to-Noise RatioCPE Customer Premises EquipmentCPW CoPlanar WaveCRABS Cellular Radio Access to Broadband ServicesDAVIC Digital Audio VIdeo CouncilDBS Direct Broadcast SatelliteDECT Digital Enhanced Cordless TelephoneDQPSK Differential Quaternary Phase Shift KeyingDRO D.. Running OscillatorDTH Direct-To-HomeDVB Digital Video BroadcastERC European Radiocommunications CommitteeETS European Telecommunications StandardETSI European Telecommunications Standards InstituteFDD Frequency Division DuplexFDDI Fibre ...InterfaceFDM Frequency Division MultiplexFDMA Frequency Division Multiple AccessGMSK Gaussian Minimum Shift KeyingGW GateWayHBT Hetrojunction Bipolar TransistorHEMT High Electron Mobility TransistorHH HouseHoldHFR Hybrid Fibre RadioIF Intermediate FrequencyIDU InDoor UnitIHDN In-House Distribution NetworkIP Internet ProtocolIRD Integrated Receiver DecoderISDN Integrated Services Digital NetworkISM Industrial, Scientific and MedicalISO International Standardisation OrganisationISP Internet Service ProviderITU International Telecommunications UnionLAN Local Area NetworkLED Light Emitting Diode


vi

LISSY Local network Interconnection using Satellite SYstemsLMDS Local Multipoint Distribution SystemLNA Low Noise AmplifierLOS Line-Of-SightMAC Medium Access ControlMAPON Master Antenna Passive Optical Network architectureMCC Main Co-ordination CentreMMDS Multi-channel Multi-point Distribution SystemMMIC Millimetre-wave Monolithic Integrated CircuitMPEG Motion Pictures Expert GroupMPT Ministry of Post and Telecommunications(UK)MVDS Multipoint Video Distribution SystemMWS Multimedia Wireless SystemNIU Network Interface UnitNRE Non Recurring EngineeringODU OutDoor UnitOFDM Orthogonal Frequency Division MultiplexPABX Private Automatic Branch eXchangePAR Peak-to-Average RatioPDH Plesiochronous Digital HierarchyPHY PHYsical layerPON Passive Optical NetworkPOTS Plain Old Telephone ServicePSTN Public Switched Telephone NetworkQoS Quality of ServiceQAM Quadrature Amplitude ModulationQPSK Quaternary Phase-Shift KeyingRF Radio FrequencyRLAN Radio Local Area NetworkSDH Synchronous Digital HierarchySME Small Medium EnterpriseSOHO Small Office/Home OfficeSONET Synchronous Optical NETworkSTB Set-Top BoxTCP Transport Control ProtocolTDM Time Division MultiplexingTDMA Time Division Multiple AccessTS Transport StreamTV TeleVisionUMTS Universal Mobile Telephone SystemVBR Variable BitRateVoD Video on DemandVSAT Very Small Aperture AntennaWAN Wide Area NetworkWDM Wave Division MultiplexingWRC World Radio ConferanceWWW World Wide Web


1

1. INTRODUCTION

The main objective of this deliverable is to provide a specification of the next generation of LMDS(Local Multipoint Distribution System) systems with focus on deployment in a pan-European market. Inthis context next generation system is understood as being commercially available within 1 to 2 yearsfrom the finalisation of the project. Furthermore, this is also intended to be the final report on theresearch activities conducted within Working Group 2. As such this document will not resemble asystem specification in the strict sense but merely reporting recommendations and guidelines forpractical and economic system design.

The emerging arenas for broadband access in Europe are currently characterised by a ongoingderegulation of the traditional market segments for services. This implies that the traditional division ofmarkets into telecom, datacom and broadcast services is breaking down and are gradually replaced by amultiple service scenario in a strongly competitive marketplace. Broadband access networks are goingdigital and consequently a diverse range of both wireline and wireless technologies will be capable ofcovering a wide range of services. As a consequence of this trend operators of broadband accessnetworks are also in position to target a broader range of user segments in the market.

This collapse of the traditional market structure is often referred to as C4 convergence and constitutes aconsiderable challenge for all players in the market of broadband access. To provide profitableoperation in this environment, the operators of access networks must in general be able to operate at lowtake-up rates and with rapidly changing service requirements. To be successful under thesecircumstances the demand for a highly flexible, scaleable and cost-effective access technology isobvious. In the context described above broadband wireless access (BWA) networks offer a verypromising/exciting alternative which is expected to fulfil most of the above mentioned requirements.

In terms of customers, the major objective is to specify a digital LMDS system suitable for deploymentin the residential and small-to-medium enterprise (SME) markets. The main source of revenue frombroadband access in the residential has traditionally been broadcasting of TV programs.Therefore, the provision of digital MPEG-2 based TV is a mandatory requirement and consequently thewell established DVB standards are chosen as a initial framework for system design. Furthermore, high-speed Internet is believed to become a mandatory requirement for the SMEs and as well as forresidential customers in the near future. A further increase of this service bouquet is envisaged and toenable a flexible handling of a highly dynamic service mix , ATM has been chosen as the basis for theswitching functionality of the system

To be able to offer both telecom and broadcasting services within the same band, there is an obviousadvantage from a regulatory point of view that fixed services (FS) and broadcasting services ( BS) areallocated on a coprimary basis in a given frequency band This requirement in fulfilled in the band 40.5-42.5 GHz which is target band for the CRABS project.


2

2. SYSTEM REQUIREMENTS AND DESIGN OBJECTIVES

2.1 Evolution of MVDS and LMDS in an European perspective

During the first half of the 1990s, the concept of Multipoint Video Distribution Services (MVDS) forcellular television was being discussed actively in Europe. The CEPT organisation had early identifiedthe band 40.5 to 42.5 GHz (as well as lower frequency bands) for such services which lead topublication of the CEPT recommendation TR/52-01 [1] published in 1992. This recommendation hasby the end of November 98 been ratified by the majority of CEPT member countries. At that timeMVDS where mostly considered for analogue TV distribution systems, with a possibly limited elementof interactivity for services such a video on demand (VoD). The standard for analogue MVDS in UKcalled MPT 1550 [3] emerged already in 1993.

In parallel the standardisation of systems for digital TV distribution in Europe progressed rapidlyheaded by the EBU body DVB and a standard for digital TV distribution via satellites [2] emerged in1994. This standard was quickly adopted by satellite broadcasters and satellite operators in thetransport part of the distribution network because it reduced the distribution cost per TV programdramatically. As the price of DVB compliant set-top-boxes (STB) dropped it became evident that thiswould pave the way for a digital MVDS standard for cellular TV distribution based on DVB standards.This would allow more TV channels, transparent interconnection of the transport and access networkand reuse of the DVB compliant STBs altogether an low-cost approach suitable for targeting theresidential market. Besides, the rapid growth in commercial Internet services at that time demonstrated afuture demand for an interactive system. Recognising that, UK Radio Agency made a revision of MPT1550 based on an interactive digital approach named MPT 1560 [4] published in June 1996. Thisstandard was later with minor adjustments adopted by ETSI as the first European standard for digitalinteractive MVDS systems [5].

At this stage the band 40.5-42.5 GHz still was allocated for broadcast services (BS) by ITU-R withboth terrestrial broadcasting and satellite broadcasting on a coprimary basis. This was changed byWRC-97 where fixed services (FS) was allocated on a coprimary basis with the other two [6] Thisdecision prepared the band for interactive services from a regulatory point of view and as such enablesthe convergence of telecom, datacom and broadcast services within a single band. This intention wasfurther enhanced by the CEPT proposal late this year which recommends this band for multimediawireless systems (MWS) including MVDS.

Considering the significant change in the use of the 40.5-42.5 GHz band, the term MVDS do no longerconvey an adequate description of the actual services expected to be provided by a next generationsystem. In the US the chosen acronym for wireless BWA systems with more or less the same servicecapability, is LMDS. Initially launched in 28 GHz band in the US, these systems are still applicable inthe 40 GHz range. Consequently, ETSI has adopted this term and a standard for the interactive pathwas issued this year [7]. In general, LMDS is now in a broader context understood by ETSI as a digitaltwo-way BWA system above 10 GHz with an asymmetric transmission bit rate. Hence, in the followingthe term LMDS will be used for the purpose of specifying the next generation system in the chosenfrequency band.

2.2 Target LMDS user classes and segments

When specifying the target services for the next generation LMDS system, one should keep in mind thepossible deployment scenarios for LMDS systems. In general the regulatory environment today ischaracterised by deregulation of the traditional market structure. This development will generate ahighly competitive marketplace where different technologies partly will deliver the same services.Consequently, the LMDS systems must be capable of delivering service quality (and availability) with acompetitive price-quality( and availability) relationship, covering a broad range of regulatory


3

100 10

1

.0 0

.1B

USI

NES

SR

ESID

ENTI

AL

Downlink Capacity [Mb/s]

Uplink C

apacity [M

b/s]SME

CORP.

0.1 1.0 10 100

HH

SOHO

Figure 1 Segmentation into LMDS user classes

regimes. These would range from a single service up to a full multimedia service configuration.Furthermore, various combinations of user classes and geographic segments will be encountered leavingthe system with a very complex and partly contradictory set of requirements.

2.2.1 User classes

The users are traditionally divided into two categories, namely business and residential users. Businessusers are different from residential users in many aspects. The major part of business customers aredemanding symmetric connections whilest the major part of the residential users are characterised byvery asymmetric demand from a broadband services point of view. Another important feature is thelarge difference in investment and tariff payment capabilites.

There is a variation within each category and to obtain a more useful representation of the customers,each user category is divided into two classes. Business users are split into corporate users (CORP.)and small- and medium-sized enterprises (SME) whilest residential users are divided between small-office and home-office (SOHO) and private households (HH). A graphical representation of the typicalmaximum uplink and downlink capacities requested by these user classes are shown in figure 1.

2.2.1.1 Corporations (large businesses)

Corporations have large demand for capacity ranging up symmetric 155 Mb/s connections andoccasionally above. These connections usually carry aggregate traffic with low burstiness factor oftenused for interconnection of LANs and PABXs by use of leased lines. A new an interesting application ishigh-speed connection to Internet.

2.2.1.2 Small- and medium-sized enterprises (SME)

SMEs are defined as businesses with up to 250 employees. The majority of businesses throughout theEuropean market are SMEs and as such they constitute an important group of customers. Due to theconsiderable number of employees the connections carry aggregate traffic typically up to symmetric E1


4

(2 Mb/s) and E2 (8 Mb/s) capacities mostly by leased lines. The high-end of this class will or hasalready migrated to higher capacities like symmetric E3 (34 Mb/s) or STM-1 (155 Mb/s).

2.2.1.3 Small-office and home-office users (SOHO)In practice, the SOHO class contains both residential and business users but has more in common withthe residential users when considering the traffic characteristics. Due to the low number of individualsinvolved, the burstiness factor is relatively high and switched connections are preferred instead of leasedlines. To cover several simultaneous services including at least multi-party video conferenceing and webbrowsing, the maximum connection capacity should be in at least 768 kb/s uplink and 2 Mb/s downlink.

2.2.1.4 Private households (HH)

The connection capacity required by private households are to a large extent determined by entertain-ment services. Reception of broadcast digital TV requires typical 6 Mb/s downlink to ensure qualitywhich compares with analogue PAL transmission. In uplink direction, the expected demand for mid-quality video conferencing or simultaneous connections of various type, requires a maximum connectioncapacity in order of 384 kb/s.

2.2.2 Geographic user segmentation

In devising a optimum system design, knowledge about local demographics is required. Due to largevariations between different countries, it is impossible to make a simple model covering a pan-Europeandeployment scenario However, to be able to suggest an efficient cellular structure (including cell size)for a given environment, a simple model must be established.

A widely used approach is to classify the users according to the average density and range for a givenarea. The average is calculated per cell area in km2 and the range is include to give a first impression ofthe corresponding variance. A segmentation model covering European households is shown in Table 1

Table 1 Typical Household (HH) densities in Europe

One should keep in mind that the local peak density could be very large compared to the average value.There are no well defined boundaries between the proposed demographic segments and particularly thetransition from suburban to urban area depends very much on local conditions.

Similar tables could be made for SOHO and SME user classes where densities in general would bemuch lower that that of private households. The SOHO user would display similarities to the HH usersin the sense that home office part would mostly be located in private households. Throughout

Demogr. segment Rural Suburban Urban City centreHH density mean 100 1000 3000HH density range 5 - 500 500 - 3000 1000 - 8000 8000 - 30,000


5

0.01 0.05 0.1 0.2 0.5 1 2 3 5 8 100

500

1000

1500

2000

2500

3000

3500

4000

HH

's pe

r c

ell

HH Density [1000/km 2 ]

LM DS cell size

0.5 km 2

1.0 km 2

2.0 km 2

4.0 km 2

8.0 km 2

25 km 2

Figure 2 Relationship between HH density and number ofHH per cell with different cell sizes

Europe the 99 % of all enterprises have less the 250 employees [8] and consequently belong to the SMEand SOHO classes. Per country basis the number of enterprises amounts to 5-10 % of the total numberof private households.

The aggregate amount of traffic generated within a single cell, is partly determined by the number ofusers connected. The number of users are related to user class density by the inverse of the cell size insquare kilometres. To predict the approximate cell size range related to the different demographicsegments, the number of HH per cell is plotted against HH density for cell size areas from 0.5 km2 to 50km2. Assuming a square cell configuration with 0.5, 1, 2, 4, 8 and 25 km2 cell size, the correspondingmaximum ranges (or diagonal of the square) are approximately 1, 1.5, 2, 3, 4 and 7 km.

If we assume a maximum number of HH per cell of 2000 the results in Figure 2 suggests that for therural segment a cell maximum. range of 3 km and up is adequate. For the suburban case between 1 and3 km is appropriate, whilest the urban case gives 0.7 to 2 km. Using the mean densities given in Table 1we arrive at 6, 2 and 0.9 km as typical average cell maximum. ranges. As already pointed out, these areaverage figures and large local variation must be expected.

2.3 LMDS service definitions

Specifying a target service scenario for the next generation of LMDS systems, one should keep in mindthat the services as such are in general not specific to a LMDS system. Rather the specific features ofLMDS could be the service attributes by which the actual services are delivered to the customer.Service attributes being employed in the following simplistic approach are uplink and downlink peakbitrate, traffic class and network/service availability.

The bitrates are specified according to established standards like fractional E1 (N*64 kb/s), E1 (2Mb/s), E2 (8 Mb/s) and E3 (34 Mb/s). This is done only for the matter of compatibility with existingequipment and does preclude the use of other bitrates. The traffic classes options are defined


6

Service Classes Uplink (kb/s) Downlink NetworkMean Peak Mean Peak Traffic Avail.

Low grade 64 64 64 64 CBR 99.99%Video Conf. Medium grade 384 384 2.048 2.048 CBR 99.99%

High grade 2.048 2.048 8.448 8.448 CBR 99.99%On Demand 16 64 3.200 4.096 VBR 99.99%

Video/TV Broadcasting - - 8.192 8.448 CBR 99.9%Contribution 8192 8192 64 64 CBR 99.99%IP-based 16 64 16 64 VBR 99.9%

Voice PSTN/ISDN 64 64 64 64 CBR 99.99%PABX Interconnect 2.048 2.048 2.048 2.048 CBR 99.99%Low grade 8 64 64 512 ABR 99%

Web Medium grade 64 512 512 2.048 VBR 99.9%High grade 512 2.048 2.048 8.192 CBR 99.99%Cellular backhaul 2.048 2.048 2.048 2.048 CBR 99.99%

Transport LAN Interconnect 8.448 8.448 8.8448 8.448 CBR 99.99%ATM UNI 25.6 25.600 25.600 25.600 25.600 V-CBR 99.99%TV-multiplex distrib. - - 34.368 34.368 CBR 99.9%

Table 2 Service classification by link capacity and network characteristics

as packet-oriented (VBR and ABR) and circuit-oriented (CBR) referring to the established terminologyfor ATM multiplexing. The last attribute is network availability which refers to the percentage of a yearfor which the network is should be available for delivery of services with the specified performance.This do not guarantee availability of the service as such because it depends on traffic loading conditionsetc.

2.3.1 Classification of target services

2.3.1.1 Video services

Video services are expected to become a very important feature of the next generation broadbandscenario. Broadband systems are in general foreseen to enable the transition from voice to video basedcommunications between individuals and between producers and consumers as well. Most videoservices are circuit-oriented and require high network availability except for TV broadcasting whichusually operate with medium availability (99.9 %). Video conferencing is a typical case wheredifferentiated service quality is expected. Hence, a tree service grade levels is proposed to cover thedifferent user classes and configurations. For the medium and high grade options multi-party videoconferencing is suggested in which asymmetric peak bit rates must be accommodated.

2.3.1.2 Voice services

Voice related service have been and will be important part of the next generation LMDS servicescenario. In addition to traditional PSTN/ISDN based telephony for communication betweenindividuals, a low to medium grade option is becoming available through IP-based telephony.PSTN/ISDN will occasionally require extremely high network availability (above 99.995 %) whichusually would not be achievable for next generation LMDS systems. From a market point of view therealso seems to be a considerable demand for aggregate voice traffic for instance through interconnectionof PABXs.


7

Service Classes HouseHold SoHo SMELow High Low High

Low Qual. 1 1 0 0 0Video Conf. Mid Qual 0 2 2 1 0

High Qual. 0 0 1 2 2On Demand 0 1 2 2 1

Video/TV Broadcasting 2 2 1 0 0Contribution 0 0 0 1 2IP-based 0 1 1 2 2

Voice PSTN/BRI-ISDN 2 2 2 1 0PABX Interconnect 0 0 0 1 2Low grade 2 1 0 0 0

Web Medium grade 1 2 2 1 0High grade 0 0 1 2 2Cellular backhaul 0 0 0 1 2

Transport LAN Interconnect 0 0 1 2 2ATM UNI 25.6 0 0 0 1 2TV-multiplex distri 0 0 0 0 1

Table 3 Typical user class service configuration according toprimary(2), secondary(1) and optional (0) services

2.3.1.3 Web browsing and Internet access

The far most important service for the next generation LMDS system is believed to be access to Internetand the world wide web (WWW). To cater for the expected diversity in service requirement, threeservice grade levels are proposed. They are differentiated in bit rate, traffic class and networkavailability. The bit rate peak-to-average ratio (PAR) is proposed to be largest for low and partly themedium grade service reflecting a presence of low interactivity users. The high grade users will be moreinteractive and may request broadband streaming capability and large file transfers both uplink anddownlink. Hence, the bit rate PAR is generally lower and the asymmetry between uplink and downlinkwould be weaker than in the case of a low or medium grade service.

2.3.1.4 Transport networking

The inherent broadband capabilities in a next generation LMDS system, makes it suitable for transportnetwork of for instance narrowband traffic. Cellular backhaul (or feeder networks) for GSM or futureUMTS networks is a evident service and interconnection of LANs is another. These services arecharacterised by high network availability and symmetric connections using a circuit-oriented approach.In case of satellite TV broadcasting, the possibility of redistributing a full MPEG-2 multiplexedconnection containing for instance 4 to 8 digital TV channels is an interesting opportunity. This requiresa downlink only and could enable a seamless integration of broadcasting into the next generation LMDSsystem.

2.3.2 User class service configurationThe different types of users specified in section

2.2.1 User classes, will request a different configurationof the services defined in Table 2. To be able to confine the LMDS system design for an adequate set ofuser classes and segments, a target service configuration for each user class must be defined. Aproposal for service configuration of the defined user classes are shown Table 3. The indicatednumbering suggests a ranking in primary(2), secondary(1) and optional(0) services. This does not


8

preclude the operation of other service configurations but merely constitutes a framework for optimisedLMDS system design.

2.3.2.1 Small and medium enterprises (SME)

SMEs are characterised by a group of people performing some kind of production activity whichusually generates a significant revenue stream. Their major use of services relates to a professionalenvironment and they will therefore mainly request high grade broadband services. SMEs are expectedto be the major user of interactive broadband video service in both contribution, video on demand andmulti-party video conferencing. They will have modern computer facilities and will probably be themajor user of IP-telephony due to their demand for long distance phone calls. Due to their size they alsomay require transport network facilities There is still a monopoly in most countries on symmetricbroadband services (E1 and above). Consequently, a large potential for leased line replacement isenvisaged.

2.3.2.2 Small Office HomeOffice (SOHO)SOHOs have much in common with the low-end of the SME user class. A professional element is beingpresent either as a stand-alone business or as residential home office connected to a SME or CORP.Consequently, one may expect medium or secondary high grade services being requested and Internetaccess and video conferencing will be important. IP-telephony will be used if long distance phone callsare vital. SOHO users also have much in common with the high end of the HH user class.

2.3.2.3 Private households (HH)

There are tree major services for private households. They are PSTN/ISDN telephony, low and mediumgrade Internet and TV broadcasting. The latter have been the major source of revenue for broadbandaccess in this user class so far. And that will probably not change within the timeframe of the nextgeneration LMDS systems. The high-end users will have more data-centric service requirements addingvideo conferencing as a primary service.

2.3.3 Definition of network availabilityThe purpose of define network availability is to provide a design objective for specification of theLMDS radio up- and downlink. From a radio point of view this figure relates to time percentage of anaverage year in which the radio link carrier-to-noise ratio (CNR) exceeds a fixed design figure.Inversely, the outage time is defined as 100 % minus availability shown the time of a average year inwhich connection is not available. A summary of different availabilitys and corresponding outage timesare given in Table 4

Grade Availability Outage timeLow 99 % 3d 15h 40minMedium 99.9 % 8h 46minHigh 99.99 % 52minExtreme 99.999 % 5min

Table 4 Availability and corresponding outage time inpercentage and time of an average year


9

Figure 3 A generic system Reference Model for InteractiveSystems

2.4 LMDS terminal configurations

2.4.1 IDU and ODU terminal requirements

User terminals for the next generation of LMDS system will in general be divided into a outdoor unit(ODU) and a indoor unit (IDU). The ODU will consist of a roof-mounted antenna closely integratedwith the microwave components necessary to do filtering, amplification and frequency translation ofsignals both for in the uplink and downlink direction whilest the IDU performs the baseband functionsrelated to uplink and downlink transmission.

The ODU should be one piece of equipment to enable simple installation at the customers premises. andthe physical interface should be at a L-band intermediate frequency (IF). This would enable possible re-use of IDU equipment manufactured for other frequency bands and access technologies. The IDUshould in principle have a common design for at least LMDS frequency bands above 20 GHz.According to the ETSI reference model [9] shown in Figure 3, IDU should logically consist of anetwork interface unit (NIU) and set-top-unit (STU). In DVB terminology the IDU is referred to as aset-top-box (STB) and another often used term is residential or commercial gateway (GW) most oftenused with reference to the business users. Preferably the STB should consist of a single piece ofequipment but that does not seem feasible for all service configurations. envisaged.

2.4.2 User classes terminal options

To devise a single/generic user terminal suitable for the whole range of user classes and serviceconfigurations solution is not regarded as a viable solution for the next generation LMDS system. Ingeneral a LMDS system tailored to a specific service configuration provides a poorer cost-performancerelationship when used for largely different service configurations. Especially, when narrowbandservices only is delivered by a true broadband system, one should usually expect to find better way ofproviding delivery. However, the possibility to add narrowband service onto broadband services, couldcreate a competitive narrowband case if accepted from a regulatory point of view.


10

B A S ES T A T I O N

C U S T O M E RU N I T

D V B

W W LD E C T

H I G HD A T ARATEA T M

M P E GC O D I N G

L O W R A T ER E T U R N

P A T H

I S D N U S I N GT D M A / F D Dfor 64 kb i t /s

channe l s

low cost un i tsw i th N x ISDNC H A N N E L S

Capac i tyequ iva lentto 2 Mbi t /s

o r more

Pr iceappropr ia te to

S M E s

interact ivelink

2 Mb i t / s

Desktop computer

Voice

Desktop computer

Ethe

rnet

Server

T V & S T B

L A N

Fax

34 Mb i t / sdown l ink

Figure 4 LMDS user terminal options.From top and downwards isshown the low, medium and high interactivity configuration

Considering the user class service configurations proposed in Table 3 there is tree obvious targetterminal options as shown in Figure 4.

The low interactivity terminal covering the low-end of the HH class, which are mainly requestingentertainment oriented service based on TV broadcasting and low grade Internet service. This approachis compliant with the architecture adopted in the ETSI reference model in Figure 3. and requires thatuplink is provide through PSTN or ISDN. The medium interactivity terminal will cover the high-endHH, SOHO and parts of the low-end SME users. Both uplink and downlink would typically requirefrom fractional to full symmetric E1 capability. And finally, the high interactivity terminal coveringmost of the SME user class requesting up to symmetric E2 capability. The medium and highinteractivity terminals both provide an integrated radio uplink with true broadband capabilities.

In terms of service configuration and terminal architecture the medium and high interactivity terminalshave much in common. Furthermore, introduction of broadband services in the market is expected totake place in the business user class. Hence, especially SMEs will be important in the first phase oflaunching next generation LMDS systems. Later as the terminal cost goes down, SOHO and high-endHH users will be included. Consequently, a common terminal for medium and high interacitivity usersare proposed for the next generation LMDS system. The low interactivity users should be served with aseparate hybrid solution with a broadband radio downlink and uplink provided by a narrowband accesstechnology.


11

3. LMDS SYSTEM ARCHITECTURE

3.1 Overview of LMDS system architecture

A LMDS system is in the following understood as consisting of all network elements necessary todeliver broadband services to users within a LMDS franchise. Hence, the LMDS system does notdirectly deal with provision of services. The choice of a network centric approach follows directly fromthe interpretation of the acronym LMDS as a generic delivery mechanism for local services. And assuch the LMDS system must to a large extent be based on a service independent architecture and avoidservice specific elements as far as possible. An example of a LMDS system architecture is shown inFigure 5.

From a network point of view, a LMDS system will in general have a four layer structure. The firstlayer originated/starts from the

LMDS control centre which is the broadband access point (BAP)connecting the LMDS system to external networks. This node in the core network is usually theentrance point for service provision by for instance ISPs. The LMDS Control Centre is connectedthrough an internal transport network to the LMDS main co-ordination centre (MCC) both whichoccasionally may be located at the same place.

The LMDS main co-ordination centre defines the transition to the second layer and is the networkentrance point to the actual geographical area in which services are delivered by the LMDS system. Atthis level, satellite broadcast TV may be inserted to lower the demand for transport network capacitybetween the control and main co-ordination centres. An overlaid transport network originates from themain co-ordination centre providing connectivity with/to the different cells within the LMDS franchise.It will provide intra-cell connectivity and the practical implementation is very much dependent on theparticular deployment scenario. The main co-ordination centre are usually located at one of the basestations at the edge of the LMDS franchise in which the cell are defined as a master cell.

Each cell is feed by a LMDS base station which defines the transition to the third level. The main co-ordination centre is connected to the base stations through the overlaid transport network. A single basestation may feed several cells and provides connectivity to a large number of user network terminationsusing radio communications in the 40 GHz band. This connectivity may be obtained by directcommunication or indirectly by use of transparent radio repeaters if necessary. This third network levelof the LMDS system is hereafter defined as the LMDS access network or LMDS network for short.This complies both with the interactive reference model in Figure 3 and the general reference modeladopted by the ETSI BRAN project [8].

To be able to provide the different services at the physical location where the users want to connect, anin-house distribution network (IHDN) is required. This part of the LMDS system defines the fourthnetwork level. The actual implementation of IHDN network may be done using for instance radio,coaxial or optical cables/technologies. In supporting multi-user dwelling units in a multi-servicescenario, the design of a cost effective and practical IHDN network is of vital importance. Generally,this level starts at the NIU as shown in Figure 3.

Although the LMDS system is described as having four network layers, there will in practice be caseswhere a transparent transition between the different levels are envisaged. This is for instance done whenemploying hybrid fibre radio (HFR) technologies at level two to feed the base stations. From a networkpoint of view, the layer structure will collapse into a simpler structure which may be advantageous froma economic point of view.


12

cell clusters inurban areas

multichanneldigital satellite

broadcastreceiver

main co-ordination

centre

trunk telecommunicationstraffic eg telephone,

internet, videoconferencing ,

SME leased lines, etc

larger cell inrural area with

near omnidirectional

antenna

micro cell

fibre/coaxconnection to main

cell clusters

radio connectionto more remote

cells

LMDS basestations

" m u s h r o o m s h a p e "

Fig .3 : Sc h em e o f th e mu sh ro o m an te nn a.

D

Figure 5 Overview of LMDS system architecture

3.2 Internal transport networks

3.2.1 Connection to external networks

Connection to external networks goes through the LMDS Control Centre. If the Control Centre and theMain Co-ordination Centre are located at the same place, the external networks are feed directly by theLMDS base station feeder network. Conversely, the traffic concentrated at the Main Co-ordinationCentre would be relayed to the Control Centre using a high capacity trunk connection. An exceptionmay be experienced if we consider an external satellite broadcasting network where a direct reception atMain Co-ordination Centre would relax the capacity requirement on the inter-connection of the MainCo-ordination Centre and the Control Centre.

The choice of transport network technology for this part of the LMDS system depends on severalfactors. First of all, the demographic characteristics of the area will impose an upper limit on therequired aggregate transport capacity. In areas with high traffic volume, a Control Centre will usuallybe established within reasonable distance and consequently a fibre optical connection would often bepreferred. If the distance to the Control Centre is large, a connection using point-to-point radio may beadequate when the traffic volume is not exceeding the transport capability of radio. This constitutes atypical case in the early stage of deployment and will be applicable for low number of customers in


13

general regardless of the actual demographic characteristics. In practice radio is expected to offer aeconomic solution in the range STM-1 (155 Mb/s) to STM-4 (622 Mb/s). For the case of broadcastonly or isolated cells far away from the Control Centre, a satellite connection may be adequatedepending on the particular LMDS cell service configuration

3.2.2 LMDS base station feeder networkThe LMDS base station feeder network corresponds to level two as defined in section

3.1 Overview ofLMDS system architecture. To deliver services provided by external networks, each LMDS cell must beconnected to Main Control Centre via a BS feeder network. In a multi-cell deployment scenario, there isa vast range of possible configuration for feeder network design. To devise an optimised solution,detailed knowledge about the expected service configuration profile for the customers within eachLMDS cell must be available.

Considering a multi-cell LMDS network, the preferred approach is to connect the LMDS cells to theMain Co-ordination Centre using an overlaid inter-cell transport network. This network areimplemented by connecting the cells in a hierarchical manner. The BS at the Main Co-ordination Centreis connected to one or more base stations which in turn again are connected to one or more base stationson the next level. In general, a nested tree structure is produced when repeating this procedure. In somecases, this structure is referred to a master-slave configuration. However, the base stations mayestablish both inter-cell and intra-cell communication independently, effectively bypassing the BS at theMain Co-ordination Centre. Thus, the master-slave terminology is not recommended in the general case.

Due to the short distance between the Main Co-ordination Centre and the nearest(master) BS, thisconnection will usually be implemented using either coaxial or optical fibre technologies. As for theinter-cell network, a microwave radio or optical fibre technologies are usually preferred. In principle,both alternatives may be used to implement a physically transparent inter-cell network in which theLMDS access network and the BS feeder network only differ in the characteristics of the transportmedium. Generally, the transition between the inter-cell and LMDS access network is non-transparenteither with different physical transmission formats or with regenerating base stations to allow fordistributed network functionality like local switching.

Three options for inter-cell network implementation are shown in Figure 6. The upper and middlealternatives are both based on microwave radio and show transparent (upper) and non-transparent(middle) implementations/configuration. In the transparent case the downlink signal in the leftmost cellis received by off-air pickup by the adjacent cells. Consequently, the same information content isdelivered in each cell. The LMDS cell uplinks are concentrated at the BS and relayed to the same cellfrom which the downlink was received. Optionally, the uplink may be provided by wireline alternativesif low uplink capacity is requested. The non-transparent alternative employs a separate microwave inter-cell backbone where the transport stream may be reformatted and remultiplexed before entering the nextnetwork level. The lower alternative shows the optical fibre option which may be transparent as well asnon-transparent. A typical transparent option is the emerging HFR technologies where all digitaloperations are performed at the Main Co-ordination Centre and the base stations only act as passiverepeaters. Non-transparent implementations are very similar to the microwave radio alternative andSDH connections carrying ATM in different configurations are typically used.


14

Figure 6 Options for inter-cell networking in a multi-cellLMDS network configuration

(a) (b)

Figure 7 BS feeder network topologies for continueous andsystematic radio coverage

To provide a continueous radio coverage throughout a wide area, a systematic procedure for designingthe BS feeder network must be employed. The choice of network topology and the technological solutionare here determined by the actual service configuration and traffic volume. Assuming a regular four cellclustering around a single BS, two possible systematic feeder network topologies are proposed in.Figure 7. Figure 7a shows a daisy-chain configuration where consecutive base stations are arranged inparallel paths with duplex transport connections. This topology is suitable for implementation oftransparent feeder networks and especially for the fibre optical solution were civil works cost can beshared when one optical fibre are feeding each BS. We also see that the uplink and downlink feeder linkpaths map into the same route. If a non-transparent implementation is preferred, this can be done usinga single high-capacity SDH ring along each chain. An alternative solution is shown in Figure 7b whichis suitable for non-transparent implementation especially for SDH microwave feeder networks. Theuplink and downlink feeder link paths here map onto several different routes and consequently feedernetwork route diversity is an inherent feature. Note the maximum number of simplex feeder link


15

connections at each BS is four and the total number is lower than for the daisy-chain configuration. Thisconfiguration may be implemented using several overlaid microwave SDH rings. Due to the largenumber of branches in the topology, this configuration is in general not suitable for fibre opticalimplementation.

3.2.3 Microwave BS feeder networks

3.2.3.1 Broadcast services delivered by off-air pickupThe first cell cluster would normally be fed through a fibre/coax connection. However, it might bedesirable to feed the next cell cluster or even an individual cell in the system, through off-air pick-up ofthe broadcast channels, until it is economic or timely to install a fibre/coax link . Assuming the four cellclustering configuration in Figure 7, the distance between two adjacent base stations would be twice theinternal cell distance. Choosing a maximum radio path length of 2km, two neighbouring base stationswould be separated by at least 4 km. We need to have a line of sight between them as shown in Figure 6and the required outage for broadcast services is about 99.9 %. The additional free space loss fromdoubling the range is 6dB and the increased rain margin of 7.5 dB in rain zone L. If a typical customerantenna with 35 dB gain was deployed to perform the off-air pick-up, the overall margin on thedownlink would be reduced by 11.5 dB in case of rain zone L.

The rainfall margin in zone L must according to Figure 13 be 25 dB for 2 km radio path length at 99.99% availability and 44 dB at 4 km. At 4 km, 99.9 % availability requires 17 dB rainfall margin. If thefirst cell cluster provide an availability of 99.99 %, a broadcasting channel availability of 99.9 % mayby achieved directly at the next cluster because the new margin is 25-6 = 19 dB which is 2 dB above therequirement of 17 dB. To achieve 99.99 % at 4 km, an additional margin of 44-25+6 = 25 dB must beprovided which could only be reached with a 35+25 = 60 dB gain antenna. In terms of practical size ofthe antenna, a availability of 99.95 % would be more adequate. It is possible that the separationsbetween the base stations may be more than 4 km with a reduction in the availability achieved in thesecond cluster. For instance, a separation of 8 km would increase the rain margin required to provide a99.9 % service to 32 dB. An antenna with a gain of 54 dB would be needed for the off-air pick-upreceiver to provide 32 dB margin.

In principle the broadcast signals could be cascaded to a further cell clusters through a tandemconnections. However the availability drops in each hop of the cascade. The availability achieved in thenth hop is a1*a2*.*an , ai is the availability achieved in the ith cell. If the availability on each hop is say99.95 % then the fourth cell in the cascade cannot be better than 99.8 %, i.e. still close to the broadcastrequirement. It is likely that the second cell cluster would be treated in much the same way as a microcell. The off-air signal would be amplified and re-transmitted on the orthogonal polarisation in thesecond cell cluster, which would use a near omni-directional antenna for this initial service for thebroadcast channels. This method could be used, even if broadcast on demand had been implemented,provided all the services demanded were also transmitted in the first cell cluster. As all the servicesrequired in the network would also be available in the first cell cluster, some break even point would bereach, where the demands for spectrum from interactive services in the initial cell cluster could renderthis arrangement inefficient and the fibre connection desirable to connect the second cluster.

3.2.3.2 Interactive services delivered over ATM

In a mature system the interactive ATM based traffic would be connected to the Main Co-ordinationCentre through a fibre link. However, in the initial stages of development connections between basestations might be achieved, as with the broadcast channels, through a radio link, using a transparentrepeater at the second cell cluster. The problem is similar but not identical to the broadcast on demandsituation. Again, the initial cluster is connected to the MCC with a fibre/coax link and a radioconnection is installed between the first and second cluster. One method of relaying services would be toassume that all the users were in the same cell cluster and relay all the services both clusters, with a line


16

of sight(l-o-s) link operating in orthogonal polarisation, as previously described using off-air pick-up.Some reduction in the availability (99.95 %) would be experienced but this would not detract muchfrom the benefits of a speedy roll-out of service achieved through a radio connection. The accessprotocols would need to accommodate the delays produced by longer path lengths of up to about 10 km(if the base station separation is 8 km).

A more complex but spectrally efficient procedure would be to split the operations between the first andsecond clusters. Then the services for the second cell cluster are carried to the next cells in a separatedfrequency band which allows the first cluster to have access to all the allocated interactive channels. Theservices would now be connected through a conventional LOS duplex link as shown in Figure 6 (i.e. nolonger be relayed through off-air pick-up). Higher gain antennas at the first cell cluster would beinstalled improving the availability achieved. At the second cell cluster, services would be re-radiatedthrough transparent repeaters on orthogonal polarisation. Initially this approach would be successful,until the demands of the second cell reach saturation in terms of the band width required on the backbone. The four cell cluster arrangement would require four time the traffic capacity(i.e. bandwidth) onthe backbone as that in each cell. Furthermore, full potential of the frequency reuse arrangements couldnot be realised with this method. All these methods use transparent repeaters at the second andsubsequent base stations, which retransmit the same modulation schemes used on the air interface forthe in-cell operations.

A further improvement in the radio connection efficiency could be achieved through using higher levelmodulation schemes on the backbone. If the modulation scheme on the backbone achieved 4 bits/Hz interms of spectral efficiency (i.e. 64 QAM), then equal bandwidth would be allocated to the backboneand four cell cluster. However this arrangement would need both regenerative repeaters at the secondbase station and a greater fade margin (12 dB more) to achieve the same availability. This complexitywould need to be compared with a fibre/coax connection in terms of cost, time scale for installation andperformance. If regenerative repeaters are used for the inter-cell connections, then one possibility wouldbe to use out-of-band frequencies for the backbone network. This then becomes a regulatory issue and itmight be considered unacceptable for a service which had already been allocated at least 2 GHz ofspectrum and possibly 3 GHz. Any encroachment on other parts of the spectrum would be prohibiteduntil a very strong case for the request could be made. However once regenerative repeaters are used forthe connections between base stations, then using equipment developed (at 38 GHz with a G.703interface) for mobile base station interconnection could be attractive in cost terms

3.2.3.3 Capacity and link margins for microwave radio backboneSome examples of the achieved performances for backbone link connections are shown below in Table 5as a function of path length, antenna gain, climatic zone and modulation scheme. The RF parametersused are similar to those for the base station systems. The bit rate is assumed to be a standard 34Mbit/s, RF power 17 dBm and noise figure 7 dB. Table 5 shows the margin achieved with theseparameters for link lengths from 4 to 10 km and compares these values with those required to achieve99.99 % availability in climatic zone E, H, and L. In zone E the 99.99 % service is just achieved onpaths up to 10 km. This is in contrast with the performance in zone L, where only the 4 km QPSKscheme meets the design goal.


17

Modulation Radio path Antenna Achieved Margin required (dB)Scheme length (km) Gain (dB) Margin( dB) Zone E Zone H Zone L

4 50 20 27 48QPSK 6 46 28 37 65

8 45 44 36 47 8110 42 43 55 94

16-QAM 8 38 36 47 8164-QAM 8 32 36 47 81

Table 5 Achieved and required margins for 99.99%microwave backbone availability

The estimation of the traffic requirements (in terms of total bit rates per cell) indicate that a fullyoperational cell might generate up to 100 Mbit/s in the busy hour. In a four cell cluster this wouldquadruple. If the backbone to cell interface is a transparent repeater, then 400 MHz bandwidth would berequired to carry backbone the traffic on a one bit/Hz basis. This seems excessive as only 600 MHz isthen left for broadcast services. A practical limit might be to assume that the backbone never carriesmore than 155 Mbit/s, i.e. about the limit of the proposed interactive spectrum allocation. In thisarrangement the backbone links would always occupy the orthogonal polarisation to that used in thecell. Some interference to users who fell within the l-o-s beam would be inevitable due to the muchhigher power densities and through scatter into the orthogonal polarisation. However the percentage ofaffected users would be small, as both the beam width of the l-o-s links is small (~100 m at 8 km range)and local terrain and building would add further protection. When the number of users in a particularcell cluster reaches about one third full capacity, that seems an opportune time to switch to a fibreconnection for the backbone.

The situation becomes more complex if the radio backbone extends to multiple hops. It would bedifficult to use much more than the above 155 Mbit/s on the backbone for interference reasons, astransmissions on the backbone on orthogonal polarisation would interfere with the in-cell services. Thusthe full capacity of all the feeder links would be limited to 155 Mbit/s for all practical purposes.


18

3.3 Cellular LMDS access network

Base station

Radio repeater

Base station

Macro cell Micro cell

Figure 8 Basic LMDS network configuration

3.3.1 The concept of a LMDS cellRadio systems at 40 GHz have two main features. Their main advantages are the very large bandwidthwhich is currently available and the small physical size of high gain antennas. The bandwidth availableof 2-3 GHz can provide nearly three orders of magnitude more capacity when compared with existingcellular mobile systems. This advantage is however qualified by the very limited range (a fewkilometres) over which such systems can provide an effective service, due to propagation effects andblockage considerations. This second restriction confines the system design to a cellular approach, whencoverage of anything larger than a very small area is considered.

The basic LMDS network configuration is shown in Figure 8. A radio cell is in this context defined as acontinuous geographical area throughout which the LMDS network will provide broadband accessaccording to specified quality and performance criteria covering a certain percentage of the potentialcustomers. To provide a suitable framework for LMDS cell design, a geographical area must becharacterised in terms of local topography, demography and meteorology. Combining the geographicalmodel with principles for radio network design, we may derive a radio cellular structure where each cellprimarily is described by its size (in square kilometre) and shape. Among the factors determining cellsize and shape, we shall later discuss the relationship to

cellular subscriber coverage, single-cellnetwork capacity and access network availability

In providing cellular broadband access, three major network elements are necessary for flexible radiocell design. Those are the cell base stations. micro-cell radio repeaters and customer premisesequipment (CPE) as shown in Figure 8. The base stations connects the cell to the internal LMDStransport network and provide connections to the individual CPEs. Usually the base stations are locatedin cell centre or at the cell edge depending each particular scenario. Note that there might be severalbase stations covering a single cell. CPEs provide the LMDS network termination at the customerlocation. To provide connection to areas without LOS to the base stations, micro-cell radio repeatersmight be necessary. The maximum range of the micro-cells is typically 0.1 to 1.0 km whilest the macro-cells illuminated by the base stations cover 1 to 10 km.


19

0 1000 2000 3000 4000 5000

Range from transmitter (m)

0

10

20

30

40

50

60

70

80

90

100%

Co

ver

age

+0m:Signal OK +0m:Signal obstructed by trees +1m:Signal OK +1m:Signal obstructed by trees +2m:Signal OK +2m:Signal obstructed by trees +3m:Signal OK +3m:Signal obstructed by trees +4m:Signal OK +4m:Signal obstructed by trees +5m:Signal OK +5m:Signal obstructed by trees

Figure 9 Radio coverage in typical north-European suburban environment as function ofdistance between BS and user.The different curves show the antenna height above the rooftop

3.3.2 Cellular subscriber coverage

In general radio systems at 40 GHz relies on line-of-sight (LOS) conditions to ensure proper operation.Hence, radio transmission at this frequency will be blocked by buildings, vegetation and terrainintercepting the radio wave path between the base station and the CPEs. This problem. constitutes oneof the major challenges of radio planning at 40 GHz. Providing 100 % subscriber coverage within thecell area may be achievable from a technical point of view but will in general not be a viable solutiondue to the substantial economic cost involved. However, this does not preclude 100 % coverage incertain areas with favourable local conditions.

3.3.2.1 Cellular design requirements and limitations

The practical limitations of subscriber coverage, is not necessarily a major problem. In the emergingcompetitive market for broadband access, different access technologies will partially target the sameservices and geographic areas. Hence, the general/total broadband access subscriber coverage would inthe long term approach 100 %. From a regulatory point of view it should then be suitable to adopt amoderate requirement on minimum subscriber coverage for LMDS networks. This would allowoperators to lower the cost of network deployment significantly enabling low-cost delivery of services tothe subscribers. To achieve a fair market penetration in a competitive scenario, the cellular subscribercoverage should be as high as possible. It should also compare with other broadband access networks interms of subscriber coverage. Consequently, a minimum subscriber coverage above 50 % is necessaryand for the purpose of LMDS cell design a subscriber coverage of 70 % is suggested.

Individual CPE LOS conditions are largely dependent on the radio path elevation angle and the height ofthe CPE antenna. For a given BS antenna height we easily see that the elevation angle decreases withincrease radio path length. Thus, for constant CPE heights, the subscriber coverage are non-uniformthroughout the cell, decreasing from the BS and outwards. This can be partly be compensated byincreasing the CPE antenna heights as you move away from the BS. BS antenna heights are estimated torange 20 to 40 m and may be restricted by local environment restrictions. CPE antennas could be easyand cheap to mount and heights up to 2-3 m above the roof line and should be manageable forresidential customers. For business customer CPE heights could be much higher at least 5 m or abovethe top of the building. There is also a matter of visual user acceptance of roof-top installations whichmay limit at the CPE antenna heights in the case of residential customers.


20

3.3.2.2 A simulation study for a typical LMDS deployment caseA typical example of a simulation of the effect of obstructions is shown in Figure 9, where a threedimensional data base, containing building, terrain and trees, has been used in conjunction with raytracing methods. The percentage of buildings, which would receive a LOS signal from a BS antenna at30 m height, is shown as a function of height of the user antenna above the roof level (upper set ofcurves). A second set of curves (lower set) shows the percentage of paths which pass through a tree.The simulation was produced for a fairly typical small town (Malvern, UK), which is located in rollingterrain with a predominance of two and three storey buildings. According to these characteristicsMalvern belongs to the overlapping section between a suburban and urban case. This is a typicaldeployment case where LMDS systems are expected to perform well.

From Figure 9 we see that the general comments concerning subscriber coverage stated previously areconfirmed. To met the design requirement of 70 % subscriber coverage, we see the maximum cell radiopath length is between 1.5 to 2.5 km. Assuming a CPE antenna height of 3 m above the roof line, thecorresponding figure is 2km. This establishes a upper bound on cell size fed by a single BS. If a edge-fed approach is used, the upper bound is equal to the maximum radio path length. However, if a centre-fed configuration is assumed, the upper bound becomes twice the maximum radio path length.Conversely, one may obtain much better subscriber coverage with a centre-fed approach (comparedwith a edge-fed) when operating with the same cell size and shape. For this particular case we alsonotice that the obstruction by trees are very low. Therefore, the major obstruction mechanism would bebuildings or other man-made objects.

3.3.2.3 Cellular network implementation guidelines

A centre-fed approach generally leads to a 7-frequency re-use pattern which in this context is a poorutilisation of the radio spectrum. Hence, this approach is suitable for rural areas where low trafficdemand allows use of larger cells and less radio spectrum. Alternatively, if an isolated cell is planned,full frequency re-use is possible and deployment in urban and suburban areas is feasible as well. Inrural areas, the blockage are mainly due to vegetation and terrain and improved coverage or enlargedcells are usually achieved by illuminating an edge-fed cell with several base stations. In case of verydifficult terrain formations (e.g. highly rolling terrain, narrow valleys), micro-cell radio repeaters maybe necessary to provide subscriber coverage. In urban to city centre environments the blockage ispredominately due to buildings and extensive use of radio repeaters in combination with multiple basestations would be needed to reach high subscriber coverage figures.

3.3.3 Cellular network capacity

3.3.3.1 Limitations on aggregate user bitrate

The maximum available LMDS network capacity is primarily determined by 3 factors. They areavailable spectrum bandwidth, spectrum re-use and spectral efficiency. Available radio spectrum at 40GHz amounts to 2 GHz with current allocations ranging from 40.5-42.5 GHz. Utilising bothpolarisations, the unable spectrum from network point of view is 4.0 GHz. There is a possibility thatan additional 1 GHz might be added later extending the band up to 43.5 GHz. In practice if a multiplecell deployment is envisaged, some kind of spectrum re-use has to be done to avoid interference betweencells. The degree of re-use is defined by the frequency or spectrum re-use factor and for a LMDSnetwork values in the range 1 to 7 will cover most deployment scenarios. A re-use factor of


21

0 0.5 1 1.5 2 2.5 3 3.5 4 4.50

2

4

6

8

10

12

14

16

S pec trum re-us e

S pec tra l E ffic ienc y [b /s /H z ]

LMD

S c

ell

ca

pac

ity [G

b/s

]

= 1= 2= 4= 7

Figure 10 LMDS cell capacity for 2 GHz of radio spectrum using both polarisations

0 0 .5 1 1 .5 2 2 .5 3 3 .5 4 4 .50

200

400

600

800

100 0

1200

1400

1600

1800

2000

Ac

tive

us

ers

per

ce

ll

S p ec t ra l E ffic ienc y [b /s /H z ]

A ve rage c apac it y pe r us e r

= 2 M b /s= 4 M b /s= 8 M b /s= 16 M b /s= 51 M b /s

Figure 11 Simultaneous number of active users per cellwith frequency re-use factor of 4 and 2 GHz of spectrum


22

one is obtained if only an isolated cell is established and a re-use factor of 7 corresponds to the case ofcentre-fed cells similar to the approach employed by mobile cellular networks. The spectral efficiencyrefers to the density with which information is carried over the radio channel. It is measured in terms ofbit per second per Hz (b/s/Hz) and is primarily defined by the coding and modulation formats used. Theaverage spectral efficiency for a LMDS network is estimated to be in the range 1.0 to 4.0 b/s/Hz. A barchart showing the relationship between LMDS network capacity and spectral efficiency for 4 practicalspectrum re-use factors is shown in Figure 10.

From the Figure 10 we see that the capacity per cell reach as high as 16 Gb/s for isolated cells. Hence,the available radio spectrum is capable of supporting for instance two duplex 2.5 Gb/s STM-16transport network connections for high density local traffic. In a multi-cell deployment scenario, afrequency re-use factor of 2 to 4 is likely to be obtained for the urban and suburban case. Assuming anaverage spectral efficiency of 1 to 2 b/s/Hz, the corresponding cell capacity will be in the range 1 to 4Gb/s being able of supporting for instance two duplex 622 Mb/s STM-4 transport network connections.For the rural case the cell capacity will be below 1 Gb/s in which one or more duplex 155 Mb/s STM-1connections may be supported.

3.3.3.2 Limitations on number of simultaneous active usersThe number of simultaneous active users handled by the LMDS network is an important design figure.Referring to the proposed user class configurations in Table 3, it is evident that no simple procedureexists for derivation of the number simultaneous users support by the LMDS network. In general, therewill be an exhaustive number of possible ways to configure a cell in terms of users and services. Toestablish an understanding of how the major factors influence the number of simultaneous active users,an approach based on a probable range of average user bitrates may be adopted. This approach will notproduce the definite answer but merely indicate the range over which the number of active users usuallyvary. A bar chart showing the relationship between spectral efficiency and number of active user foraverage user bitrates in the range 2 to 51 Mb/s, is shown in Figure 11. The assumptions are a frequencyre-use factor of 4 and 2 GHz of available spectrum.

With the stated assumptions, the bar chart shows that the number of simultaneous users may reach ashigh as 2000. In the range 1 to 2 b/s/Hz which seem to be the most likely option in practice, the figureis below 1000. According to the user class service configurations in Table 3, an average duplex userbitrate of 2 Mb/s may be interpreted as a border between the business and residential segment. Hence,we may state that a practical LMDS network may support at least 1000 HHs or up to 1000 SME percell. The typical average duplex user bitrates will in practice be much lower than 2 Mb/s for HHs andhigher than 2 Mb/s for SMEs. Consequently, as a first order conclusion we may suggest that the LMDSnetwork will typically support up to several 100 SMEs or several 1000 HH per cell. In a mixedresidential and business scenario, the figures would have to be reduced. Also, if for instance half thespectrum is used for broadcasting of digital TV, the corresponding numbers should be reduced with afactor of two. Considering a multi-licence scenario, a similar reduction may be envisaged.

One should notice that the upper bounds of simultaneous active users does not restrict the deployment ofLMDS networks in terms of customer or HH densities. The mapping between the number of activeusers and customer densities requires knowledge about radio coverage, service penetration the useractivity rate (that is the ratio between active and total number of customers)


23

Figure 12 Climatic zones in Europe according to ITU-RP.837[11]

3.3.4 Access network availability

As stated in section 2.3.3 Definition of network availability, the access network availability is in thiscontext based on the yearly CNR statistics as determined by radio events. This constitutes a frameworkfor predicting a best-case lower bound of LMDS network availability. and other processes havingimpact on network availability (like MAC protocols), will not be capable of improving this lowerbound.

There are several radio propagation phenomena contributing to a reduction of the nominal CNR. The byfar most important one in terms of reduced signal level, is rainfall attenuation. Rainfall attenuation ischaracterised by large and slow variations and the fading process is generally not considered to befrequency selective over the bandwidth of a LMDS up- or downlink. Annual predictions models forrainfall attenuation based on [12] show that its scales strongly with rain rate and radio path length. Thelatter shows that there is a relationship between rain attenuation statistics and maximum radio pathlength within LMDS cell and furthermore the corresponding cell size.

The major meteorological parameter employed in calculating the attenuation prediction is point rainrate.Ideally, accurate knowledge about local rainrate statistics is required to derive correct predictions.However, due to lack of such data in most places, the ITU-R recommendation P.837 [11] is used as ageneral framework for the predictions. ITU-R P.837 prescribes a division of the earth in climatic zoneswith largely similar rainrate statistics. Recommended climatic zones in Europe is depicted in Figure 12.The map shows that the Mediterranean area along with Turkey, parts of Eastern Europe and theBalkans occasionally encounter very intense rainfall (zone K and L).


24

05

10152025303540

0,0 0,5 1,0 1,5 2,0 2,5 3,0 3,5 4,0 4,5 5,0

0

5

10

15

20

25

30

35

40

EFGHJKL

05

10152025303540

(a)

10.1.1.200.3302.pdf

Documents

generation lmds system

generation systems

generation of lmds equipment

cellular radio access

lmds radio subsystems

title of deliverable

system architecture

residential market