RF PLANNING .pdf

7/23/2019 RF PLANNING .pdf

1/80

Number/Version Checked by Approved by Page

1.0.0 23 Oct 98 JRy 1(80)

Frequency Hopping Network Planning Guide


2/80


Version Date Page

1.0.0 Oct 23, 1998 2/80

HISTORY

Version Date Author Comments

0.0.1 21 Sep, 1998 MaSa The first draft

0.0.2 24 Sep, 1998 JRy Modifications: The whole document restructured,

Chapter 2.3: PC and DTX gains, Chapter 7.2: RXQual

distribution, Table 9: Ho Threshold Interference.

Added: Figure 5-13, Figure 5-14, Figure 7-3, Figure 7-4,

Figure 7-8, Table 10, Table 11, Table 12.

Added: History , Chapter 2.1.6, Chapter 3.6, Chapter

3.9.1, Chapter 3.9.6, Chapter 3.9.7, Chapter 3.9.8,

Chapter 5.1, Chapter 5.2, Chapter 5.6.2, Chapter 6.3,

Chapter 7.1, Chapter 7.2, Chapter 7.9.

1.0.0 23 Oct, 1998 JRy The first accepted version


3/80


Version Date Page

1.0.0 Oct 23, 1998 3/80

CONTENTS

1. INTRODUCTION...........................................................................................................................5

1.1 GENERAL DESCRIPTION OF FREQUENCY HOPPING ...........................................................................5

1.2 FREQUENCY HOPPING MODES ........................................................................................................61.3 CELL ALLOCATION ........................................................................................................................8

1.4 MOBILE ALLOCATION ....................................................................................................................9

1.5 HOPPING SEQUENCE NUMBER ........................................................................................................9

1.6 MOBILE ALLOCATION INDEX OFFSET .............................................................................................9

1.7 MAIO STEP.................................................................................................................................10

2. THEORETICAL PERFORMANCE OF FREQUENCY HOPPING.........................................11

2.1 FREQUENCY DIVERSITY ...............................................................................................................11

2.1.1 Coherence Bandwidth..........................................................................................................11

2.1.2 Effect of Interleaving ...........................................................................................................13

2.1.3 Cyclic vs. Random Hopping Sequences................................................................................14

2.1.4 Simulated Frequency Diversity Gains............. .....................................................................14

2.1.5 Effect in Cell Coverage Area ...............................................................................................16

2.1.6 Effect of Mobile Speed.........................................................................................................16

2.2 INTERFERENCE DIVERSITY ...........................................................................................................16

2.3 EFFECT OF POWER CONTROL AND DTX........................................................................................18

3. NOKIAS SUPPORT FOR FREQUENCY HOPPING IN GSM................................................20

3.1 BSS LEVEL IMPLEMENTATION .....................................................................................................20

3.2 THE 2ND GENERATION BASE STATION..........................................................................................20

3.3 TALK FAMILY BASE STATION.......................................................................................................21

3.4 PRIMESITE ...................................................................................................................................22

3.5 BASE STATION CONTROLLER........................................................................................................23

3.6 NPS/X.........................................................................................................................................23

3.7 MAXIMUM CONFIGURATIONS .......................................................................................................23

3.8 RADIO NETWORK FAULT MANAGEMENT ......................................................................................24

3.8.1 The 2nd Generation Base Station........................................................................ .................25

3.8.2 Talk Family Base Stations and PrimeSite ........................................................... .................25

3.9 RESTRICTIONS ON THE USAGE OF FH............................................................................................25

3.9.1 DL Power Control with BB FH........................................................................... .................25

3.9.2 Downlink DTX.....................................................................................................................26

3.9.3 Extended Range Cell (DE34/DF34/DG35) ..........................................................................26

3.9.4 MS Speed Detection.............................................................................................................26

3.9.5 Half Rate.............................................................................................................................26

3.9.6 Frequency Sharing ..............................................................................................................263.9.7 RTC Combiner ....................................................................................................................26

3.9.8 NPS/X..................................................................................................................................26

4. SELECTING THE RIGHT HOPPING STRATEGY.................................................................27

5. FREQUENCY PLANNING OF FREQUENCY HOPPING NETWORKS ...............................29

5.1 NETWORK PLANNING PROCEDURE................................................................................................29


4/80


Version Date Page

1.0.0 Oct 23, 1998 4/80

5.2 FREQUENCY PLANNING PROCEDURE WITH NPS/X ........................................................................30

5.3 FREQUENCY REUSE ON FREQUENCY HOPPING NETWORK ..............................................................335.3.1 Effective Reuse ....................................................................................................................34

5.3.2 Frequency Allocation Reuse (RF FH only) ..........................................................................34

5.4 LOAD ON NETWORKS UTILISING FRACTIONAL LOADING (RF FH ONLY)........................................35

5.4.1 Frequency Load.............................................. .....................................................................35

5.4.2 Hard Blocking Load............................................................................................................365.4.3 Fractional Load.............................................. .....................................................................37

5.5 TRUNKING EFFECT AND EFFECTIVE REUSE ...................................................................................38

5.6 FREQUENCY ALLOCATION STRATEGIES ........................................................................................40

5.6.1 BCCH Allocation.................................................................................................................40

5.6.2 Selecting the Effective Reuse (BB FH) .................................................................................43

5.6.3 Selecting the Frequency Allocation Reuse and the Frequency Load (RF FH).......................44

5.6.4 Frequency Sharing by Using MAIO Management (RF FH only) ..........................................46

5.6.5 Frequency Sharing in the Single MA-list Scheme (RF FH only)...........................................50

6. RADIO NETWORK PARAMETERS .........................................................................................52

6.1 PARAMETERS FOR MA-LIST DEFINITIONS IN BSC .........................................................................52

6.2 BTS LEVEL FH RELATED PARAMETERS .......................................................................................546.3 POWER CONTROL.........................................................................................................................56

6.4 HANDOVER..................................................................................................................................58

6.5 DTX............................................................................................................................................59

6.5.1 Uplink DTX.........................................................................................................................59

6.5.2 Downlink DTX.....................................................................................................................59

7. OPTIMISATION..........................................................................................................................60

7.1 TOOLS FOR NETWORK MONITORING.............................................................................................60

7.2 KPIS FOR HOPPING NETWORK ......................................................................................................60

7.3 RXQUAL IN FH NETWORKS .......................................................................................................61

7.4 IDLE CHANNEL INTERFERENCE MEASUREMENT ............................................................................65

7.5 CYCLIC AND RANDOM HOPPING SEQUENCES ................................................................................667.6 INTRACELL HANDOVER................................................................................................................69

7.7 POWER CONTROL.........................................................................................................................69

7.7.1 Downlink Power Control with BB Hopping .........................................................................70

7.8 HANDOVER CONTROL ..................................................................................................................70

7.9 HSN PLANNING WITH RANDOM HOPPING.....................................................................................70

8. PLANNING CASES .....................................................................................................................71

8.1 PLANNING CASE 1: SINGLE MA-LIST............................................................................................71

8.1.1 Frequency Planning ............................................................................................................71

8.1.2 MAIO Planning ...................................................................................................................72

8.2 PLANNING CASE 2: RF FH WITH FRACTIONAL LOADING (FAR 3 5) ...........................................75

8.2.1 Defining the Frequency Band and the Number of Frequencies Needed in Each Cell............758.2.2 Frequency Allocation and Analysis......................................................................................77

8.3 PLANNING CASE 3: RF FH WITH FREQUENCY SHARING ................................................................78

8.3.1 Frequency Planning ............................................................................................................78

8.3.2 MAIO Planning ...................................................................................................................79

8.3.3 Analysis...............................................................................................................................80


5/80


Version Date Page

1.0.0 Oct 23, 1998 5/80

1. INTRODUCTION

The purpose of this document is to explain the theory behind the frequency hopping (FH), how the

frequency hopping is implemented in Nokias network elements, how to choose the right frequency

hopping strategy, parameters related to FH, frequency allocation procedure, how to analyse the

quality of the network and the optimisation process. Also some practical planning examples arepresented.

Frequency hopping is one of the standardised capacity enhancement features in GSM system. It

offers a significant capacity gain without any costly infrastructure requirements. It is also compatible

with all the existing GSM mobile phones, since the frequency hopping support has been required by

the GSM specifications from the beginning. Frequency hopping can co-exist with most of the other

capacity enhancement features and in many cases it significantly boosts the effect of those features.

All these factors make frequency hopping a very tempting capacity enhancement solution.

Figure 1-1. Solutions to enhance network capacity.

1.1 General Description of Frequency Hopping

Frequency hopping can be briefly defined as a sequential change of carrier frequency on the radio

link between the mobile and the base station.

In GSM, one carrier frequency is divided into eight time slots. Each time slot provides one physical

channel, which can be assigned to one link between a mobile and a base station. The communication

between the mobile and the base station occurs in bursts inside the assigned time slot. Each burstlasts about 577 s. When frequency hopping is used, the carrier frequency may be changed between

each consecutive TDMA frame. This means that for each connection the change of the frequency

may happen between every burst. This is called Slow Frequency Hopping(SFH), because more than

one bit is transmitted using the same frequency. In Fast Frequency Hopping (FFH), the carrier

frequency is allowed to change more than once during a bit duration, but this is not implemented in

GSM.

CAPACITY GAIN

Effective Network Planning

Channel-Bandwidth Spectrum Cell Size Reuse-Factor (C/I)

Dual-Band-/Dual-Band-/

Dual-Mode-Dual-Mode-

NetworksNetworks

PC DTX FHPC DTX FH

Smart AntennasSmart Antennas

IUOIUOIFHIFH

Half-RateHalf-Rate

NetworksNetworksAntennas DownAntennas Down

Ant.Ant.DowntiltingDowntiltingMicro-CellMicro-Cell

Pico-Cell / IndoorPico-Cell / Indoor


6/80


Version Date Page

1.0.0 Oct 23, 1998 6/80

At first, the frequency hopping was used in military applications in order to improve the secrecy and

to make the system more robust against jamming. In cellular network, the frequency hopping also

provides some additional benefits such as frequency diversity and interference diversity. The basic

principle of frequency hopping is presented in Figure 1-2.

Frequency

Time

F1

F2

F3

Call is transmitted through several

frequencies in order to average the interference (interference diversity) minimise the impact of fading (frequency diversity)

Figure 1-2. Basic functionality of frequency hopping.

1.2 Frequency Hopping Modes

The requirement that the BCCH TRX must transmit continuously in all the time slots sets strict

limitations on how the frequency hopping can be realised in a cell. The current solutions are

Baseband Frequency Hopping(BB FH) and Synthesised Frequency Hopping(RF FH).

In the baseband frequency hoppingthe TRXs operate at fixed frequencies. Frequency hopping is

generated by switching consecutive bursts in each time slot through different TRXs according to theassigned hopping sequence. The number of frequencies to hop over is determined by the number of

TRXs. Because the first time slot of the BCCH TRX is not allowed to hop, it must be excluded from

the hopping sequence. This leads to three different hopping groups. The first group doesnt hop and it

includes only the BCCH time slot. The second group consists of the first time slots of the non-BCCH

TRXs. The third group includes time slots one through seven from every TRX. This is illustrated in

Figure 1-3.


7/80


Version Date Page

1.0.0 Oct 23, 1998 7/80

B

RTSL 0 1 2 3 4 5 6 7

TRX-1

TRX-2

TRX-3

TRX-4

f1 B= BCCH timeslot. It does not hop.

f2

f3

f4

Time slot 0 of TRX-2,-3,-4 hop over f2,f3,f4.

Time slots 1...7 of all TRXs

hop over (f1,f2,f3,f4).

Figure 1-3.Baseband hopping (BB FH).

In the synthesised frequency hoppingall the TRXs except the BCCH TRX change their frequency

for every TDMA frame according to the hopping sequence. Thus the BCCH TRX doesnt hop. The

number of frequencies to hop over is limited to 63, which is the maximum number of frequencies in

theMobile Allocation(MA) list covered in Section 1.4. Synthesised hopping is illustrated in Figure1-4.

BTRX-1

Non-BCCH TRXs are hopping over

the MA-list (f1,f2,f3,...,fn) attached to the cell.

TRX-2

B= BCCH timeslot. TRX does not hop.

f1,

f2,

f3,

fn

f1,

f2,

f3,

fn

. . . .

Figure 1-4.Synthesised hopping (RF FH).

The biggest limitation in baseband hopping is that the number of the hopping frequencies is the same

as the number of TRXs. In synthesised hopping the number of the hopping frequencies can be

anything between the number of hopping TRXs and 63. However in synthesised hopping the BCCH

TRX is left completely out of the hopping sequence. The differences between BB and RF hopping

are further illustrated in Figure 1-5.

HSN2

HSN1

HSN1


8/80


Version Date Page

1.0.0 Oct 23, 1998 8/80

MSC

BB-FHF1(+ BCCH)

F2

F3

Dig. RF

TRX-3

TRX-1

RF-FH

F1, F2,F3

Dig. RF

TRX-1

TRX-2

BSCTCSM

BCCH

Frequency

Time

F1F2F3

MS does not seeany difference

BB-FH is feasible with large configurations

RF-FH is viable with smaller configurations

Figure 1-5. The difference between BB and RF FH.

1.3 Cell Allocation

The Cell Allocation (CA) is a list of all the frequencies allocated to a cell. The CA is transmitted

regularly on the BCCH. Usually it is also included in the signaling messages that command the

mobile to start using a frequency hopping logical channel. The cell allocation may be different for

each cell.

In GSM 900 the CA list may include all the 124 available frequencies [GSM 04.08]. However, thepractical limit is 64, since the MA-list can only point to 64 frequencies that are included in the CA

list as presented in the next section. The only signaling method allowed in the GSM 900 systems to

transmit the CA list is the bit map 0 method presented in Table 1.

Table 1. The signalling method for transmitting the CA list in GSM 900 system.

CA signaling

method

Lowest

ARFCN

Max. ARFCN range Max. number of

frequencies in the CA list

bit map 0 0 124 124** Practical limit is 64, because the MA-list can only point to 64 frequencies.

In GSM 1800 and GSM 1900 systems the frequency band is so large that the CA list cannot includeall the frequencies available in a system. In these systems the bit map 0 method is not available, butfive other methods can be used [DCS 04.08] [J-STD 7]. Each of these methods has different

limitations that limit the maximum frequency range and the maximum number of frequencies. These

signaling methods together with their limitations are presented in Table 2. In Nokia implementation

the variable bit map and the 512 range signaling methods are available. The CA list is always

automatically generated and it includes the BCCH frequency and the frequencies that are defined for

the MA-list.


9/80


Version Date Page

1.0.0 Oct 23, 1998 9/80

Table 2. Different signalling methods for transmitting the CA list in GSM 1800/1900.

CA signaling

method

Lowest

ARFCN

Max. ARFCN

range

Max. number of

frequencies in the CA list

1024 range 0 1024 16 (17 if ARFCN 0 is included in the CA list)

512 range selectable 512 18256 range selectable 256 22

128 range selectable 128 29

variable bit map selectable 112 112** Practical limit is 64, because the MA-list can only point to 64 frequencies.

1.4 Mobile Allocation

The MA is a list of hopping frequencies transmitted to a mobile every time it is assigned to a hopping

physical channel. The MA-listis a subset of the CA list. The MA-listis automatically generated if the

baseband hopping is used. If the network utilises the RF hopping, the MA-lists have to be generated

for each cell by the network planner. The MA-listis able to point to 64 of the frequencies defined in

the CA list. However, the BCCH frequency is also included in the CA list, so the practical maximum

number of frequencies in the MA-list is 63. The frequencies in the MA-list are required to be in

increasing order because of the type of signaling used to transfer the MA-list.

1.5 Hopping Sequence Number

The Hopping Sequence Number (HSN) indicates which hopping sequence of the 64 available is

selected. The hopping sequence determines the order in which the frequencies in the MA-listare to be

used. The HSNs 1 - 63 are pseudo random sequences used in the random hopping while the HSN 0 is

reserved for a sequential sequence used in the cyclic hopping. The hopping sequence algorithm takes

HSN and FN as an input and the output of the hopping sequence generation is a Mobile Allocation

Index(MAI) which is a number ranging from 0 to the number of frequencies in the MA-listsubtracted

by one. The HSN is a cell specific parameter. For the baseband hopping two HSNs exists. The zero

time slots in a BB hopping cell use the HSN1 and the rest of the time slots follow the HSN2 as

presented in Figure 1-3. All the time slots in RF hopping cell follow the HSN1 as presented in Figure

1-4.

1.6 Mobile Allocation Index Offset

When there is more than one TRX in the BTS using the same MA-list the Mobile Allocation Index

Offset(MAIO) is used to ensure that each TRX uses always an unique frequency. Each hopping TRX

is allocated a different MAIO. MAIO is added to MAI when the frequency to be used is determined

from the MA-list. Example of the hopping sequence generation is presented in Figure 1-6. MAIO and

HSN are transmitted to a mobile together with the MA-list. In Nokia solution the MAIOoffsetis a cellspecific parameter defining the MAIOTRX for the first hopping TRX in a cell. The MAIOs for the

other hopping TRXs are automatically allocated according to the MAIOstep-parameter introduced in

the following section.


10/80


Version Date Page

1.0.0 Oct 23, 1998 10/80

GSM Hopping algorithm

MAI(0...N-1)=

f1 f2 f3 f4 fNfN-1MA

0 1 2 3 N-1N-2MA INDEX(MAI)

TRX-1 TRX-2 TRX-3

FN & HSN

MAIOTRXTRX-1 0

TRX-2 1TRX-3 2

For thisTDMA frame the output from the algorithm is 1

1

1

+ MAIOTRX

Figure 1-6. Example of the hopping sequence generation.

1.7 MAIO Step

The MAIOstepis a Nokia specific parameter used in the MAIO allocation to the TRXs. The MAIO for

the first hopping TRXs in each cell is defined by the cell specific MAIOoffsetparameter. MAIOs for

the other hopping TRXs are assigned by adding the MAIOstepto the MAIO of the previous hopping

TRX as presented in Equation (1.1).

)1()( += nMAIOMAIOMAIO stepoffsetnTRX (1.1)

An example of the MAIO assignment is presented in Figure 1-7. More examples can be found in

Section 5.6.4.

Sector TRX # HSN MAIO stepMAIOoffsetl MAIO

1 1 Non-hopping BCCH TRX

2 7 2 0 0

3 2

4 4


2 7 2 6 6

3 84 10


2 7 2 12 12

3 14

4 16

MAIO step indicates the

difference between the MAIOs of

successive TRXs in a cell.

+MAIO step

Figure 1-7. Example of the use of the MAIO related parameters.

MAIOOFFSET ,

User definable

These parameters

are set

automatically


11/80


Version Date Page

1.0.0 Oct 23, 1998 11/80

2. THEORETICAL PERFORMANCE OF FREQUENCY HOPPING

Frequency hopping is a powerful countermeasure in order to overcome the harmful effects introduced

by the propagation channel and interference. The quality gain achieved by employing frequency

hopping can be traded for capacity gain by tightening the frequency reuse in the network.

2.1 Frequency Diversity

The fast fading is a significant problem especially in the downlink direction since the mobiles do not

employ antenna diversity, which is commonly used in base stations. Fluctuations of the received

signal strength are especially harmful for the slow moving mobilesbecause they tend to stay in a

fading dip much longer than the faster moving mobiles. Frequency hopping causes the consecutive

bursts to be transmitted on different frequencies. If the separation between these frequencies is

sufficient, the fading characteristics of these frequencies are different.

For the fast moving mobiles, the consecutive bursts have different fading characteristics even

without frequency hopping, because the spatial movement between the consecutive bursts is

significant and the locations of the fading dips are relatively constant in most environments. Thus thefrequency diversity gain for the fast moving mobiles is not significant.

2.1.1 Coherence Bandwidth

Coherence bandwidth represents a bandwidth that is required between two frequencies in order to

ensure that their fading characteristics are different enough to provide properly uncorrelated

amplitudes and phases. The coherence bandwidth depends strongly on the mean delay spread of the

environment.

Because of the multipath scattering, the transmitted impulse signal spreads in time domain before it is

received. A typical signal delay envelope of a transmitted impulse is presented in Figure 2-1. Theparameters as defined in [Lee82] are

d tE t dt m =

( )0

(2.1)

2 2

0

2=

t E t dt d m( ) , (2.2)

where:

dm= mean excess delay time

t= excess delay time

E( )= signal power density

= delay spread


12/80


Version Date Page

1.0.0 Oct 23, 1998 12/80

0

0 dB

dMean delay time

tDelay time

Powerdensity

E(t)

Delay spread

Figure 2-1. Typical delay envelope.

The delay spread is thus defined as the standard deviation of the mean delay time. The measurements

indicate that the delay spread is highly dependent on the environment. Typical values are presented in

Table 3 [Lee89].

Table 3. Mean delay spreads

Type of environment Delay spread , s

Open area < 0.2

Suburban area 0.5

Urban area 3

The coherence bandwidth is often defined as the frequency separation that yields an autocorrelation

coefficient value of 0.5 or less [Pen95]. If the propagation environment is also time dependent, the

time separation of signals has to be taken into account. The autocorrelation coefficient based on the

frequency and time separation can be written as follows [Lee82]

r

J v( , )

( )

( )

=

+

0

2

2 21

, ( 2.3 )

where

J0( ) = Bessel function of 0th order

= 2/, = signal wavelength

v= velocity of the mobile

= time separation

= delay spread of the environment

= 2*f, f= frequency spacing

Adequate coherence bandwidth, where signal autocorrelation coefficient between bursts equals to

0.5, can be derived from Equation (2.3) assuming = 0 as

BWC( . )

= =

0 51

2 . ( 2.4 )


13/80


Version Date Page

1.0.0 Oct 23, 1998 13/80

Equation (2.4) can be fully applied only in an ideal case, and it is therefore only a theoretical model.

However, it gives an idea about how the coherence bandwidth differs in different types of

environments. In Figure 2-2 the autocorrelation coefficient has been plotted for several different

values of delay spread () assuming = 0. It can be seen that in the urban environmenteven theadjacent channelhaving separationof 200 kHzappears to be adequately uncorrelated and in the

suburban environmentthe channel separation of 400 kHzis adequate. In open environmentsthe

channel separation should be at least 800 kHzcorresponding to four GSM carriers.

0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1.0

0

100

200

300

400

500

600

700

800

900

1000

1100

1200

1300

1400

1500

frequency spacing (kHz)

a

utocorrelationcoefficient

0.2

0.5

1

2

3

delay

spread ( s)

Figure 2-2. The autocorrelation coefficient as a function of carrier spacing.

2.1.2 Effect of Interleaving

In GSM the speech frame is transmitted over eight consecutive bursts. The fast fading causes bursty

bit errors that degrade the efficiency of the convolutional coding. The interleaving is designed to

spread these errors over longer time. However, the decoding performance is not significantly

improved if consecutive bursts are exposed to the similar radio channel. If the mobile moves fast

enough, the fading of successive bursts is uncorrelated due to spatial movement. Frequency hopping

causes consecutive bursts to be transmitted on different frequencies. If these frequencies have

sufficient separation the fading of successive bursts is uncorrelated as presented in Section 2.1.1.

Since the interleaving depth is eight, the frequency diversity gain of cyclic hopping doesnt

significantly improve if more than eight frequencies are used in a hopping sequence.

In data calls, the interleaving length is 19. Therefore, the gain for data calls compared to speech

calls might be bigger when more than 8 frequencies are used in a hopping sequence.

The signalling channels have an interleaving depth of four. The frequency diversity gain for the

signalling channels is thus smaller.


14/80


Version Date Page

1.0.0 Oct 23, 1998 14/80

2.1.3 Cyclic vs. Random Hopping Sequences

Both cyclic and random hopping modes are available in GSM.

In the cyclicmode the frequencies are changed sequentially from the lowest frequency to thehighest as defined in the MA-list.

In random mode the frequency to be used for each burst is selected from the MA-list by a

predefined pseudo random sequence. This means that the same frequency may be used for acouple of consecutive bursts and the frequencies are not used evenly in a short time scale.

Thus, the optimum frequency diversity gain is possible to achieve only if the cyclic hopping is

used. As the number of frequencies becomes larger the difference between the cyclic and the random

mode becomes small.

2.1.4 Simulated Frequency Diversity Gains

0

1

2

3

4

5

6

7

8

9

10

No hop 2 3 4 5 6 8 Infinite

Number of carriers

Eb/N0(dB)

FLAT 3FER = 3%

TU3FER = 3%

FLAT3RBERCl 1b = 0,3%

TU3RBERCl 1b = 0,3%

Figure 2-3. Frequency diversity gain of frequency hopping link against thermal noise

compared to a non-hopping link.


15/80


Version Date Page

1.0.0 Oct 23, 1998 15/80

0

1

2

3

4

5

6

7

8

9

No hop 2 3 4 5 6 8 Infinite

Number of carriers

C/Ic(dB)

FLAT 3

FER = 3%

FLAT3RBERCl 1b = 0,2%

TU3FER = 3%

TU3RBERCl 1b = 0,2%

Figure 2-4. Frequency diversity gain of frequency hopping link against co-channel

interference compared to a non-hopping link.

The simulations show a very significant gain for FLAT3 channel compared to the TU3 channel. This

happens because the TU3 channel includes several propagation paths having statistically independent

fading conditions and it is thus providing path diversity that helps to achieve the performance targets

even in the non-hopping case. The results of this simulation represent a best possible case, because

the fading on the used frequency channels is assumed uncorrelated and the cyclic hopping mode is

used. In real life, the frequencies are not necessarily uncorrelated as explained in Section 2.1.1 and

the random hopping is used to maximise the interference diversity gain . Also, the presented

gains are not achievable in uplink direction if a proper diversity reception (about 4 dB gain) method

is already in use at base stations.

According to the simulations, the performance of the SACCH / SDCCH and TCH for the cases of

non hopping and ideal FH as a function of C/I (according to 05.05 test conditions and TU3) are

presented in the following:

Table 4. The frequency diversity gain of the SACCH / SDCCH against TCH for the

cases of non hopping and ideal FH as a function of C/I, with 2%FER.

TCH/FS SACCHNo FH 15dB 11.5dB

FH 8dB 8dB

In the non hopping mode, the SACCH is more robust than the TCH/FS, whereas in the FH mode they

perform equal.


16/80


Version Date Page

1.0.0 Oct 23, 1998 16/80

2.1.5 Effect in Cell Coverage Area

In coverage limited cells the frequency hopping may increase the cell coverage areabecause of

the frequency diversity gain, but since the BCCH time slot doesnt hop, the increased coverage area

is relevant only for the ongoing calls that have been successfully established and are allocated a

hopping TCH. According to the simulations, see Table 4, the non-hopping signalling channel (BCCH

/ SDCCH) has a better performance than a non-hopping TCH but a worse performance than ahopping TCH channel. Therefore, the cell coverage area could be increased, but not according to

the full FH gain, but by considering the performance of the BCCH time slot.

In RF FHcase, the whole BCCH carrier is non-hopping. Thus, the frequency diversity gain should

be considered as a quality gainin the cell border area rather than the gain increasing the cell service

area.

2.1.6 Effect of Mobile Speed

As mentioned earlier, the frequency diversity gain for the fast moving mobiles is not significant. The

movement as itself causes the same gain which is lost from the frequency diversity gain. Therefore,the fast moving mobiles get the same gain than the slow moving ones, the gain just comes more or

less from the moving as itself.

In GSM, the speed of Power Control (PC) is slow. When moving fast, the PC cannot follow anymore

the slow fading dips so efficiently. Therefore, the fast moving mobiles might loose in PC gain. Also

the Handover (HO) performance may be degraded with high speed.

2.2 Interference Diversity

In a conventional non-hoppingnetwork, each call is transmitted on a single fixed frequency. This

means that the interference situation in a network is also quite stable. Some calls may experience

very little interference and the other calls may be interfered severely. Severe interference can beavoided by a handover, but the probability of finding an interference free channel decreases as the

network load increases. In a non-hopping network, the interference tends to be continuous, so that the

same interference source affects several consecutive bursts. If this interference is strong enough it

may lead to a corruption of several consecutive bursts. The error correction measures used in GSM

can not usually tolerate several corrupted bursts in a speech frame and thus these frames are likely to

be erased causing significant deterioration in speech quality.

In random hoppingnetwork, the interference sources vary from burst to burst. Thus, the interference

tends to get averaged over all the calls in the network. As a consequence, the interference affectingeach call in the network has a lower standard deviation around its mean value. This effect is

illustrated in Figure 2-5. Another advantage of random frequency hopping is that the severely

interfered bursts occur randomly. Because of this, the probability of several consecutive corruptedbursts and erased frames decreases.


17/80


Version Date Page

1.0.0 Oct 23, 1998 17/80

0

5

10

15

20

25

30

Call 1 Call 2 Call 3

Aver

ageC/I(dB)

0

5

10

15

20

25

30

Call 1 Call 2 Call 3

AverageC/I(dB)

Figure 2-5. Interference averaging between users in a random frequency hopping network.

In order to use the available frequency spectrum efficiently, the frequencies are reused in a network.

The sufficient distance between the cells using the same frequency depends on the minimum C/I ratio

tolerated by the system, the surrounding environment and the network topology. In practice the

minimum reuse for a non-hopping macro cells is about 12. This means that the same frequency maybe used in every 12th cell. Because the interference levels for each user vary considerably, a large

interference margin has to be included to guarantee sufficient quality for each user in the network.

When the random frequency hopping is employed the deviation of interference level is decreased as

illustrated in Figure 2-5. This means that the interference marginused in the frequency planning

can be reducedallowing the usage of tighter frequency reuse as illustrated in Figure 2-6.

Field strenght

Serving carrier

averageweakestinterference

averagestrongestinterference

interferencemargin

worstinterference

FH with tighterfrequencyreuse

FH withimprovedquality

no FH

Figure 2-6. The gain of frequency hopping.

How big is the interference diversity gain is a subject for a further study.

f3f

f2

f3

f2

f

f3

f2

f

f2

f3

f1

Ave

FHNo FH


18/80


Version Date Page

1.0.0 Oct 23, 1998 18/80

2.3 Effect of Power Control and DTX

Both the power control and the DTXare standard GSM features, which are designed to minimise

the interfering transmission when possible. They are both mandatory features in the mobile

terminals, but it is up to the network operator to decide whether to use them or not. DTX prevents

unnecessary transmissions when there is no need to transfer information. Power control is used to

optimise the transmitted signal strength so that the signal strength at the receiver is still adequate. Theboth features can be individuallyactivated for uplink and downlink. Operators have been widely

using both features in UL directionmainly in order to maximise the battery lifein mobiles.

In a non-hopping network these features provide some quality gainfor some users, but this gain

cannot be transferred effectively to increased capacity, since the maximum interference experienced

by each user is likely to remain the same. Also the power control mechanism doesnt function

optimally because the interference sources are stable causing chain effects where the increase of

transmission power of one transmitter causes worse quality in the interfered receiver, which in turn

causes the power increase in another transmitter and so on. This means that, for example, one mobile

located in a coverage limited area may severely limit the possibility of several other transmitters to

reduce their power.

In a random hopping network the quality gain provided by both features can be efficiently

exploited to capacity gainbecause the gain is more equally distributed among the users. Since the

typical speech activity factor (also called DTX factor) is less than 0.5, DTX effectively cuts the

network load in half when it is used. In a soft blocking limited network this means that the DTX can

theoretically provide up to 100% capacity increase. Also, the power control works more efficiently

because each user has many interference sources. Thus, if one interferer increases its power, the

effect on the quality of the connection is not seriously affected. In fact, it is probable that some other

interferers are decreasing their powers at the same time. Thus, the system is more stable and chaining

effects mentioned earlier do not occur frequently.

The simulated gain for power control and DTX with different mobile speeds can be seen in the

following Figure 2-7.

GAIN:PC on 1.4 dBDTX on 2.3 dBPC on, DTX on 3.7 dB

GAIN:PC on 1.0 dBDTX on 2.3 dBPC on, DTX on 3.5 dB

Reuse 3/9, TU 3km/h Reuse 3/9, TU 50km/h

C/I improvement

Figure 2-7. The simulated gain of PC and DTX with FH.


19/80


Version Date Page

1.0.0 Oct 23, 1998 19/80

DTX has some effect on the RXQual distribution. Normally the BER is averaged over the duration

of one SACCH frame lasting 0.48 seconds and consisting of 104 TDMA frames. However, four of

these TDMA frames are used for measurements, so that only 100 bursts are actually transmitted and

received. When DTX is in use and there is no speech activity, only the bursts transmitting the silence

descriptor frame (SID-frame) and the SACCH are transmitted. When there are periods of no speech

activity, the BER is estimated over just the bursts carrying the silence descriptor frame and the

SACCH. This includes only 12 bursts over which the BER is averaged (sub quality). This means thatthe BER gets averaged much more effectively when DTX is not used yielding to a quality

distribution where the proportion of moderate quality values is enhanced. The sub quality

distribution is wider than the full quality distribution, meaning that more good and bad quality

samples are experienced.

The differences between full and sub quality distributions are largest in frequency hopping networks

utilising low frequency allocation reuse, since in that kind of networks the interference situation

may be very different from burst to burst. A couple of severely interfered bursts may cause very bad

qualityfor the sub quality sample when they happen to occurin the set of 12 bursts over which the

sub quality is determined. The full quality sample of the same time period has probably only

moderate quality deterioration because of the better averaging of BER over 100 bursts. The

differences between full and sub quality distributions can be seen in Figure 2-8.

In a real network utilising DTX the quality distribution is a mixture of full and sub quality samples.

The proportions of full and sub samples depend on the speech activity factor also known as the DTX

factor. The differences in the BER averaging processes cause significant differences in the RXQUAL

distributions. These differences should be taken into account when the RXQUAL distributions of

networks utilising and not utilising DTX are compared.

1/1 reuse 15 freqs

0.00 %

5.00 %

10.00 %

15.00 %

20.00 %

25.00 %

30.00 %

35.00 %

40.00 %

Q0 Q1 Q2 Q3 Q4 Q5 Q6 Q7

RxQ full

RxQ sub

Figure 2-8. The distribution of normal RXQual and subRXQual values in a frequency

hopping network.

The limitations in the usage of DL PC and DTX can be seen in Chapters 3.9.1 and 3.9.2.


20/80


Version Date Page

1.0.0 Oct 23, 1998 20/80

3. NOKIAS SUPPORT FOR FREQUENCY HOPPING IN GSM

The support for frequency hopping is a standard feature of NokiaBase Station Sub-System(BSS). In

this chapter the frequency hopping support of different base station generations and the BSC are

described. Also the current and upcoming frequency hopping support of Nokias radio network

planning tool NPS/X is presented.

3.1 BSS Level Implementation

In GSM only the BSS is responsible of the implementation of frequency hopping. TheNetwork Sub-

System(NSS) including theMobile Switching Centre(MSC) is not involved in it. The Operation and

Maintenance Centre (OMC) is involved in managing the FH related parameters, but their

management in the OMC doesnt differ from any other cell level parameter. The fault management in

the OMC of a frequency hopping network is identical to that of a non-hopping network. The primary

network elements in GSM are presented in Figure 3-1.

MSC

MS

OMC

BSC

BTS

BTS

NSSBSS

Ainterface

Abisinterface

Figure 3-1. The primary network elements in GSM.

3.2 The 2nd Generation Base Station

The second generation base station supports only baseband hopping. The main functional blocks in

the second generation BTS considering frequency hopping are the Frame Units(FU), the Frequency

Hopping Unit(FQHU) and the Carrier Units(CU) [Nok96]. The frame unit performs all the controland the baseband functions for frames of up to 8 full rate or 16 half rate logical channels. Each carrier

unit contains a transmitter and two receivers. The main function of the transmitter is to convert the

digital data from the frame unit into a modulated carrier signal. The receiver is responsible for the

down conversion from the RF frequency band to baseband followed by A/D conversion and

serialising I and Q signals and sending them to the demodulation part in the corresponding frame unit

[Nok95]. The number of frame units and carrier units corresponds to the number of installed TRXs in

the BTS.


21/80


Version Date Page

1.0.0 Oct 23, 1998 21/80

The frequency hopping connects the frame units and the carrier units as illustrated in Figure 3-2. The

hopping function is realised by multiplexing baseband digital bit streams between the frame units and

the carrier units. The multiplexing is done according to the hopping sequence, which is calculated in

FQHU. The hopping unit is common for the BTS; all the sectors of a BTS use the same FQHU. The

FQHU can be duplicated for reliabilityorbecause of diversity reception. If the diversity is not

used, the other FQHU acts as a hot redundancy, which means that it is automatically taken intooperation if the other FQHU fails. When diversity reception is used, the other FQHU is used for

carrying the signal from the diversity receiver.

FU1

FU2

FU3

FU12

CU1

CU2

CU3

CU12

F

Q

H

U

Figure 3-2. Functional units for frequency hopping in 2nd generation BTS.

The FQHU is capable of supporting a maximum of 12 hopping groups at a time. This is sufficient as

in three sector configuration the number of hopping groups used is nine (including the non-hopping

zero time slots on the BCCH carriers). Both random and cyclic hopping modes are supported but not

simultaneously, meaning that all the sectors under the same BTS must use either cyclic or random

hopping sequences. With random hopping the hopping sequence numbers (1-63) can be selected

freely for each hopping group.

The timing of sectors is derived from a common clock unit, so the different sectors are frame- and

bit-synchronised enabling the use of synchronous handovers. Consequently, the hopping sequences

are synchronised as well. The combiners used in the 2nd generation BTSs limit the minimum

channel spacing to 600 kHz!

3.3 Talk Family Base Station

The Talk family base stations are capable ofboth baseband hopping and RF hopping. Baseband

hopping implementation is slightly different compared to the implementation on the 2nd generation

base station. Functionality inside one TRX is divided between burst level operations (EQDSP) and

block level operations (CHDSP). The burst level operations cover all the operations done for a singleburst, such as ciphering/deciphering, equalisation, bit detection, diversity combining etc. The block

level operations deal with blocks of information, such as a speech block or a signaling block. These

operations include interleaving/deinterleaving, block coding/decoding etc. The baseband hopping

interface resides between this logical division.


22/80


Version Date Page

1.0.0 Oct 23, 1998 22/80

FHDSP is a digital signal processor dedicated to controlling the frequency hopping operation. In

baseband hopping the FHDSP controls the information transfer between the EQDSP and the

CHDSP realising the frequency hopping as illustrated in Figure 3-3. The FBUS is a two-way parallel

bus dedicated for this purpose and dimensioned to support a maximum of 12 TRXs.

TRX1,

CHDSPTRX1,

EQDSP

F

B

U

S

FHDSP

TRX2,

EQDSP

TRX3,

EQDSP

TRX12,EQDSP

TRX2,

CHDSP

TRX3,

CHDSP

TRX12,CHDSP

Figure 3-3. Baseband hopping implementation in the Talk family base stations.

With RF hopping the FBUS is also used, but the connections are always made one-to-one. For

example, the EQDSP of TRX1 is always connected to the CHDSP of TRX1. The FBUS is then used

for sending the RF channel number from the FHDSP to be used on the next time slot. Two

synthesiser banks are used, while one is in use the other is being tuned to the frequency used in the

next time slot. Delivery of channel numbers from FBUS to synthesisers is done by hardware.

RF hopping and BB hopping cannot be used simultaneously. This means that all the sectors under

the same Base Station Control Function (BCF) must use the same hopping method, if any.

However, some sectors may be hopping while others remain non-hopping. The used combiner

type may also restrict the possibility of utilising RF hopping. IfRemote Tuned Combiners(RTC) are

used, the RF hopping cannot be used. This is because the RTC is based on tuneable cavities, which

cannot be retuned dynamically according to the used hopping sequence. The minimum channel

spacing when RTC is used is 600 kHz. The other combiner option for the Talk family base stations

is the wide bandAntenna Filter Equipment(AFE). AFE supports both BB and RF hopping and there

are no minimum channel spacing requirements.

3.4 PrimeSite

PrimeSite is a small highly integrated base station based on the Talk family technology. It contains

only one TRX and the hardware is reduced, so that the FBUS have been removed and the functions

of FHDSP have been integrated to the CHDSP.


23/80


Version Date Page

1.0.0 Oct 23, 1998 23/80

The RF hopping can be implemented by connecting two or more PrimeSites together as a multi-

TRX configuration. In this case the first PrimeSite provides the BCCH carrier and is thus in a non-

hopping mode, whereas the other connected PrimeSites are hopping according to the hopping

sequence.

The BB hoppingis also possible to arrange with the PrimeSites by using properties of RF hopping.

This pseudo-BB hopping appears outwards similar to the pure BB-hopping. Pseudo-BB hopping ispossible when two or more PrimeSites have been connected for a multi-TRX configuration. The

PrimeSite is able to transmit the first time slot (RTSL 0) by using a different frequency than the other

time slots. The pseudo-BB hopping is realised by transmitting the RTSL 0 on the BCCH TRX on one

fixed frequency and the other time slots by using a frequency determined according to the hopping

sequence. The other TRXs use the HSN1 for the RTSL 0s and HSN2 for the RTSLs 1-7 as described

in Section 1.2. The number of frequencies in the pseudo-BB hopping equals the number of connected

PrimeSites for RTSLs 1-7 and one less for the RTSL 0. A dummy signal is sent on the BCCH

frequency in the non-active TCH time slots.

3.5 Base Station Controller

The BSC functionality related to frequency hopping is implemented by software. There are no

hardware dependencies. The frequency hopping management in the BSC is quite simple. The main

principle is that the BSC is handling logical channels on the cells under its control. The logical

channels may then be assigned on the frequency hopping physical channels, but they are provided by

the base stations. The basic requirement for the BSC is to handle the additional parameters (MA,

MAIO and HSN) needed to define a hopping logical channel. The parameters are stored in the BSS

Radio Network Configuration Database (BSDATA) in the BSC, maintained by the Operation and

Maintenance(O&M) software.

The radio resource management doesnt know about frequency hopping. It allocates the logical

channels as usual. The hopping related parameters are attached later by the Abis interface program

block, which reads the needed hopping related parameters from the database. The parametersdefining a frequency hopping channel are then attached to Abisand Air interface signaling messages.

In Abisand Air interface radio resource management signaling the frequency hopping is affecting the

CHANNEL_ACTIVATION (Abis), IMMEDIATE_ASSIGNMENT (Air),

ASSIGNMENT_COMMAND (Air) and HANDOVER_COMMAND (Air) messages.

3.6 NPS/X

NPS/X is an integrated software package for the cellular network planning developed by Nokia. See

more details of the FH support and the planning and frequency allocation process in Chapters 5.1 and

5.2.

3.7 Maximum Configurations

Maximum BTS configurations are presented in Table 5.

Table 5. Maximum BTS configurations in different BSS software releases.


24/80


Version Date Page

1.0.0 Oct 23, 1998 24/80

BSS 6

BTS type Combiner type BCFs

Antennas/cell

(polarisation

diversity used) Combiner type BCFs

Antennas/cell

(polarisation

diversity used)

2nd generation RTC (BB FH)

omni 10 TRXs 1 1

sectorised 4+4+4 TRXs 1 1

Talk family RTC (BB FH) AFE (BB & RF FH)

omni 6 TRXs 1 1 4 TRXs 1 1 sectorised 6+6+6 TRXs 2 1 4+4+4 TRXs 1 1

Prime Site Standard

(BB & RF FH)

sectorised n*y TRXs 1) 1 1

BSS 7 2nd generation RTC (BB FH)

omni 10 TRXs 1 1



omni 6 TRXs 1 1 12 TRXs 1 3

sectorised 6+6+6 TRXs 2 1 12+12+12 TRXs 3 3

Prime Site Standard

(BB & RF FH) sectorised n*y TRXs 1) 1 1

BSS 8 2nd generation RTC (BB FH)

omni 10 TRXs 1 1



omni 12 TRXs 1 1 12 TRXs 1 3

sectorised 12+12+12 TRXs 2 1 12+12+12 TRXs 3 3

Prime Site Standard

(BB & RF FH)

sectorised n*y TRXs 1) 1 1

1) The amount of sectors is not limited; even each TRX can be a sector of its own. Max. 16 TRXs per BCF are allowed.

They can be freely divided into sectors of different sizes. Only rule is that n*y must be less than or equal to 16.

3.8 Radio Network Fault Management

The radio network configuration management in the BSC determines the recovery actions in

abnormal situations in the BSS radio network, such as faults, fault cancels and initialisations. The

recovery actions are executed if errors occur in the functional blocks of the BTS, such as the carrier

unit, the frame unit, the tranceiver, functional blocks common to the whole cell or the functional

blocks common to the whole BTS site. In addition to this, the recovery options are executed if the D-

channel of the Abisinterface fails or if there are failures detected by the call control of the BSC in the

connection with the radio channel allocation procedure. The recovery actions are determined based

on the type of the faulty functional block and they are based on the radio facilities configured to thefaulty block.


25/80


Version Date Page

1.0.0 Oct 23, 1998 25/80

3.8.1 The 2nd Generation Base Station

A frame unit fault may be either internal or external. The internal fault could be caused, for

example, by a FU hardware malfunction and the external fault could happen, for example, because of

a lost LAPD-link to the TRX. In both cases the recovery procedure is similar. The procedure is as

follows:

1. The BTS alarms the BSC or the BSC detects a non-functional LAPD-link.

2. The BSC clears all the calls that are allocated to those Abiscircuits corresponding to the faulty

TRX. Calls on the other TRXs proceed normally.

3. The BSC blocks the faulty frame unit in order not to allow new traffic for the A bis circuits

corresponding to it.

The calls on the other TRXs can proceed normally and the hopping parameters can be left untouched,

because all the carrier units are still functioning. The mobiles on the cell can still hop over all the

frequencies originally allocated to that cell.

In case of a carrier unit faultone tranceiver doesnt work properly. Thus, one of the frequencies in

the hopping sequence cannot be transmitted and/or received properly. In this case the procedure is asfollows:

1. The BTS alarms the BSC.

2. The BSC blocks all the TRXs of the cell for a while. This causes clearing of all the ongoing calls

on that cell.

3. The BSC calculates new hopping parameters including a new MA-listin which the frequency of the

faulty CU is removed.

4. The BSC unblocks the TRXs that have functioning CUs and the new hopping parameters are

transferred to the BTS.

5. The BSC allows new traffic for the functioning TRXs.

3.8.2 Talk Family Base Stations and PrimeSite

In a case of BB hopping the procedure is similar to the carrier unit fault in the 2nd generation BTS as

described in the previous section.

If the BTS is RF hopping, the recovery procedure is similar to the frame unit fault in the 2nd

generation BTS as described in the previous section.

3.9 Restrictions on the Usage of FH

3.9.1 DL Power Control with BB FH

In BB FH the BCCH carrier is involved in the hopping sequence. The BCCH carrier is always sent in

the downlink direction with the maximum power defined for the cell. When the PC is used in the

other than BCCH carrier, there is a big difference in the sent / received power between the carriers.

The gain control of some mobiles cannot follow so big and sudden changes in the received power.

Therefore, it is recommended to restrict the PC range in DL direction to 10-15 dB with BB FH.


26/80


Version Date Page

1.0.0 Oct 23, 1998 26/80

3.9.2 Downlink DTX

Baseband hopping combined with downlink DTX causes problems in the mobile stations, because in

the silent phase the dummy frames are sent on the BCCH frequency causing malfunction in the

mobile stations. ETSI has approved a solution to solve the problem and it is implemented in Nokia

BSS. The solution is to use a special training sequence code in the dummy burst but it does not

guarantee that all mobile station models of different manufacturers are working error free.

3.9.3 Extended Range Cell (DE34/DF34/DG35)

Only RF hopping is supported, and only for the TRXs serving the normal coverage area. The TRXs

serving the extended coverage area cannot hop.

3.9.4 MS Speed Detection

The speed detection algorithm in the BTS works only for non-hopping channels. In a case of

frequency hopping the speed information in the Measurement Result message from BTS to BSC is

set to value 'non-valid' indicating that speed information is not available from that particular cell.

3.9.5 Half Rate

The interleaving depth of the TCH/HS is four instead of eight as it is in TCH/FS. Because the

interleaving has a significant effect on the successful error correction of the speech frame, especially

on the frequency hopping networks utilising low frequency allocation reuse and fractional loading,

the performance of frequency hopping may be reduced.

The use of cyclic hopping with even number of hopping frequencies should be avoided in networks

utilising half rate. Since the half rate channel is transmitted on every other TDMA frame, the usage of

cyclic hopping with even number of frequencies means that one half rate connection uses only half of

the frequencies. This problem doesnt occur if random hopping sequences are used.

3.9.6 Frequency Sharing

The basic requirement in frequency sharing (1/1 reuse, 3/3 reuse) is that the cells at one site have to

be controlled by the same BCF, so that they are frame synchronised. With the current Nokia

equipment this requirement limits the maximum TRX configuration to 12 TRXs per site.

3.9.7 RTC Combiner

In the 2nd

generation and Talk Family base stations, the RTC combiners have the limitation of the

minimum channel spacing of 600 kHz.

3.9.8 NPS/X

NPS/X 3.2and the older versions dont support frequency allocation for a fractional loaded network

(= more frequencies than TRXs). NPS/X 3.2 can estimate the quality of the fractional loaded

frequency plan.

NPS/X 3.3can make the channel allocation for a fractional loaded network.


27/80


Version Date Page

1.0.0 Oct 23, 1998 27/80

4. SELECTING THE RIGHT HOPPING STRATEGY

The goal in the selection of the hopping strategy is to maximise the effectiveness of frequency

hopping in order to achieve a maximum capacity and/or quality gain. The basic requirement for

the maximum FH gain is to make sure that each cell has a sufficient number of frequencies in the

hopping sequence. Equally important is that a good frequency plan minimising the interference canbe produced.

The BTS hardwaremay severely restrictthe possibilities. Second generation base stations are only

capable of BB hopping. The Talk family (3rd

gen) base stations support also RF hopping, but only if

wide band combiners (AFE) are used.

The maximum TRX configurationswhich can be used with different hopping modes (combiners)

and hopping schemes (maximum TRX amounts under the same BCF) can easily become also

restricting factors.

The amount of antennasand antenna feeder cablescan be a limiting factor. With AFE combiner,

about three times more antennas are required than with RTC combiner.

The utilisation of RF hopping is preferable if downlink power controlis used. In BB hopping the

DL PC causes dramatic changes in DL field strength as some of the bursts are transmitted by the full

power BCCH TRX and the rest of the bursts by low power TRXs. The mobile receivers cannot

tolerate quick changes of field strength resulting to poor DL quality. To avoid this problem the

maximum power reduction for DL PC in conjunction with BB hopping should be limited to 10 15

dB. This limitation is likely to reduce the achievable gain from DL PC.

As in conventional network, the successful implementation of RF hopping with fractional loading

requires a good frequency plan that minimises the interference in the network. Usually the best

results can be achieved with a help of a frequency allocation tool. However, the frequency allocation

is not possible for fractionally loaded networks if the frequency allocation tool doesnt supportfractional loading. For NPS/X this support is available in version 3.3.

There is, however, one special case of RF hopping with fractional loading that doesnt require any

frequency planning at all. In this single MA-listschemeall the frequencies are allocated to every cell

so that the frequency allocation reuse is 1. In many cases this scheme may not provide the best

possible gains, but the gain compared to a non-hopping network is still significant as verified in a

trial that was conducted in a real network. If the frequency band is extremely limited, the application

of a single MA-listmay be the only sensible way to implement FH, because it always provides the

maximum number of frequencies to hop over in every cell.

Another possibility is to utilise frequency sharingarrangement. In this scheme all the cells of one

site share the same MA-listin a controlled manner so that interference between the cells of the same

site can be avoided. Frequency sharing makes it possible to have enough hopping frequencies in

every cell without a need to utilise fractional loading. Thus, the frequency planning is possible with

tools that dont support fractional loading.

The main factors affecting the decision of the frequency hopping strategy are presented in Figure 4-1.


28/80


Version Date Page

1.0.0 Oct 23, 1998 28/80

Figure 4-1. Flow chart for hopping strategy decision.

Min TRX configuration

3 TRX/cellor more

2 TRX/cell

BTS generation2nd gen.

3rd gen. only

Planning tool supportsFH and fractional loading

YesNo

Easy planning preferredover maximum capacity

Yes

No

BB FH used on the cellshaving more than 2 TRXs

max 6 TRX / cell with RTCor 12 TRX with AFE

RF FH with frequencyallocation reuse 1(=single MA list scheme)

max 12 TRX / site!(under the same BCF)

RF FH with frequencyallocation reuse 3 ~ 5

max 12 TRX / cell

Combiner type /Amount of antennas

RTC

AFE

Yes

No

Maximum gain fromDL PC required

RF FH with frequencysharing (no fractionalloading)

max 12 TRX / site!

(under the same BCF)


29/80


Version Date Page

1.0.0 Oct 23, 1998 29/80

5. FREQUENCY PLANNING OF FREQUENCY HOPPING NETWORKS

Frequency hopping requires some new considerations in the frequency planning process. This is

especially important if the RF hopping with fractional loading is used. The frequency planning of

fractionally loaded networks requires special attention to the load control. On the other hand, the RF

hopping allows some new planning concepts like frequency sharing and the control over frequencyallocation reuse while the effective reuse in the network remains the same.

Large TRX configurations make baseband hopping feasible. In order to achieve a proper

frequency hopping gain, a minimum of three TRXs in a cell should be used with the baseband

hopping [Tun97]. The benefit of the baseband hopping is that the TCHs located on the BCCH TRX

are included in the frequency hopping sequence. The BCCH frequencies have a high frequency reuse

in order to guarantee a successful signaling and a fast decoding of the base station identification code.

It is beneficial to have this interference free BCCH frequency included in the hopping sequence,

because it is likely to improve the quality of reception on the hopping logical channels.

In frequency planningpoint ofview, the planning of a baseband hoppingnetwork differs less than

the planning of a RF hopping network from the planning of a conventionalnon-hopping network.The main difference is that the fractional loading is not possible when the baseband hopping is used.

Because of this, it is possible to use the conventional frequency planning toolswhen planning the

baseband hopping network. However, because of the interference and frequency diversity gains,

lower C/I ratios and therefore smaller frequency reuse distances can be allowed in the baseband

hopping network compared to a non-hopping network.

5.1 Network Planning Procedure

The network planning and monitoring process for a baseband frequency hopping network is

basically the same than for a non-hopping network. The planning of an RF hopping networkcan

be a little more complex, if the maximum capacity is wanted to get out from the network. Thesuitable frequency allocation scheme have to be selected and the frequency load must be equalized to

guarantee an equal quality distribution.

If a tight frequency allocation scheme has been chosen then the estimation of thesubjective speech

qualitycan become a more challenging task compared to a non-hopping network. When FH is used

the RXQual distribution is not anymore comparable to the non-hopping network.

NPS/X is an integrated software package for the cellular network planning developed by Nokia. It

provides the basic tools for coverage prediction, frequency allocation and interference analysis. The

propagation modeling is based on digital maps presenting both the terrain type information and the

height data of the target area. Available propagation models include Okumura-Hata, Juul-Nyholm,

Walfish-Ikegami and a ray-tracing model. The ray-tracing model is specifically for microcellplanning and it is available in NPS/X version 3.2.

NPS/X versions before version 3.2 dont include any frequency hopping specific support . New

versions called NPS/X 3.2 and 3.3 have some new functionalities to make the frequency planning and

the quality analysis an easier task, see Chapter 5.2.


30/80


Version Date Page

1.0.0 Oct 23, 1998 30/80

Figure 5-1. Network planning and monitoring process.

Capacity PlanningCapacity Planning

Frequency PlanningFrequency Planning Parameter PlanningParameter Planning

MonitoringMonitoring

NPSXNetdim

NDW

NPSXNetdimNDW

NPS/X 3.3NPS/X 3.3

NDWNDW

PlanEditCDW

PlanEditCDW

NMS/2000NMS/2000

5.2 Frequency Planning Procedure with NPS/X

NPS/X versions before version 3.2 dont include any frequency hopping specific support. However,

since the frequency hopping doesnt affect the propagation, the coverage planning phase is not

different when planning frequency hopping networks compared to non-hopping networks. In

coverage limited cells the frequency hopping increases the cell coverage area because of the

frequency diversity gain. Since the BCCH time slot doesnt hop, the increased coverage areamust

be dimensioned according to the performance of BCCH time slotinstead of hopping TCHs, see

Chapter 2.1.5. For this reason, the frequency diversity gain should be considered as a quality gain in

the cell border area rather than a gain increasing the cell service area.

For the planning of baseband hopping networks the traditional frequency allocation andinterference analysis tools are also sufficient. Due to the frequency diversity and interference

diversity gains the hopping allows somewhat worse C/I ratios compared to a non-hopping network.

This can be taken into account when setting parameters for the frequency allocation tool leading to a

tighter frequency plan. When analysing the resulting plan, higher interference levels can be tolerated.

Frequency hopping specific planning tool support is needed when RF hopping with fractionalloading is used.Fractional loading means that a cell is allocated with more frequencies than there are

TRXs.

The quality prediction toolin NPS/X 3.2estimates the downlink RXQUAL for every pixel in the

work area. These values can be displayed in the digital map using different colours for particular

RXQUAL levels. From the map overlay the areas potentially suffering from interference can beeasily identified. To make the comparison between different plans easier, a statistics window is also

implemented. This window presents the distribution of predicted RXQUAL values in the work area.


31/80


Version Date Page

1.0.0 Oct 23, 1998 31/80

The prediction is based on the C/I ratio that is calculated by using the field strengths of the serving

carrier and the interfering carriers. The corresponding Bit Error Ratio(BER) is determined from the

calculated C/I ratio. The calculations take the DTX factor and the load factor into account where

appropriate. When the BER for the pixel is calculated it is converted to RXQUAL value according to

the mapping specified in GSM specifications [GSM 05.08]. The input parameters needed for the

calculation are the frequencies allocated for the cells, the DTX factor and the blocking probability for

each cell. Both base band and RF hopping modes are supported. Note, that the frequencyallocation for a fractional loaded network is not supported in NPS/X 3.2.

Figure 5-2. Example output from the RXQUAL prediction tool.


32/80


Version Date Page

1.0.0 Oct 23, 1998 32/80

Figure 5-3. RXQUAL statistics window.

NPS/X 3.3will include a new frequency allocation tool, which is capable of allocating frequencies

utilising low frequency reuse and fractional loading. Also the MAIO Management can be taken into a

use. The MA list lengths can be defined manually in cell basis, or NPS/X can define themautomatically by a certain criteria. After the MA list length has been chosen the allocation algorithm

tries to produce an optimal allocation. In high interfered areas longer MA list lengths can be tried to

average the interference.

Also the network simulator of NPS/X 3.3 includes a support for FH.


33/80


Version Date Page

1.0.0 Oct 23, 1998 33/80

Figure 5-4. Frequency allocation procedure.

5.3 Frequency Reuse on Frequency Hopping Network

Since the frequency band is always limited, the frequencies have to be reused in the network. As the

reuse distance becomes smaller, there are more frequencies available for each cell. Because each

TRX in a cell requires a unique frequency, the capacity potential of a cell is increased, as there are

more frequencies available for each cell. However, when the reuse distance becomes small enough,

all the frequencies available for the cell cannot be utilised because of too severe interference in the

cell border areas. For a conventional non-hopping network this is the practical frequency reuse limit.

The BB hopping network has this same limit, but because of frequency hopping gain, somewhat

lower reuse distances are allowed before the quality reaches the minimum acceptable limit.

The advantage of RF hopping is that the frequency reuse distance can be set as low as wanted. This

can be done, because a RF hopping cell can use more frequencies than there are TRXs installed. This

means that the used frequencies are only fractionally loaded as presented in Section 5.4.3. For a

fractionally loaded RF hopping network, two reuse figures have to be defined. These are effective

reuseand frequency allocation reuse. They are presented in the following sections.

Capacityestimation,cell basis

Capacityestimation,cell basis

Planningconceptdecision

Planningconceptdecision

Estimation ofneeded numberof frequencies

Estimation ofneeded numberof frequencies

Coverage dataCoverage data

Neighbour cellmeasurements with

GPA tool

Neighbour cellmeasurements with

GPA tool

InterferenceCalibration Tool

InterferenceCalibration Tool

Interference matrixgeneration

Interference matrixgeneration

Automatic interferergeneration for IUO

Automatic interferergeneration for IUO

Frequencyrequirements

Frequencyrequirements

FrequencyAllocation

FrequencyAllocation

Spectrumand HWconstraints

Spectrumand HWconstraints

NPS/X 3.3

Planning of otherparameters

Planning of otherparameters

OMC / CDW/ NDW

Quality Analysis

AutomaticParameter tuning

Quality Analysis

AutomaticParameter tuning

NetDim /NPS/X


34/80


Version Date Page

1.0.0 Oct 23, 1998 34/80

5.3.1 Effective Reuse

The effective reuse is essentially the same as the conventional frequency reuse distance. It is

calculated as

R NN

efffreqsTOT

TRXave

= , ( 5.1 )

where:

Reff= effective reuse

NfreqsTOT= total number of used frequencies

NTRXave= average number of TRXs in a cell

Since the effective reuse takes the actual number of frequencies together with the number of TRXs

into account, it can be also used as a capacity index, provided that the TRXs can be loaded at least to

the hard blocking limit as presented in Section 5.4.2. The smaller the effective reuse, the higher the

capacity in terms of the number of TCHs provided by one frequency in the network.

5.3.2 Frequency Allocation Reuse (RF FH only)

Frequency allocation reuse indicates how closely the frequencies are actually reused in a network.

Thus, it indicates the severity of a worst case C/I in the cell border. It is calculated as

FARN

N

freqsTOT

freqs MA

=/

, ( 5.2 )

where:

FAR = frequency allocation reuse NfreqsTOT= total number of used frequencies

Nfreqs/MA= average number of frequencies in MA-lists

If the network doesnt utilise fractional loading, the frequency allocation reuse is the same as the

effective reuse. Example of the reuse calculations for the fractionally loaded RF hopping network is

presented in Figure 5-5.


35/80


Version Date Page

1.0.0 Oct 23, 1998 35/80

Frequency Allocation Reuse = Total # offrequencies / # of frequencies in MAL

Effective Reuse = Total # of frequencies/Number of TRXs per cell

Frequency Allocation Reuse Effective Reuse

Total # of freqs = 30

4 TRXs / cell

10 frequencies / cell

3

21

1 12 2

33

1/3

FAR = 30/10 = 3

Eff.reuse = 30/4 =7.5

Example:Example:

Figure 5-5. Example of reuse calculations.

5.4 Load on Networks Utilising Fractional Loading (RF FH only)

One of the most essential parameters of the fractionally loaded RF hopping network is the load. The

load on the frequencies is the most important one since it determines the probability of collisions.

Collision means that the serving cell and an interfering cell are transmitting at the same frequency at

the same time so that the potential interference becomes reality.

5.4.1 Frequency Load

When designing a network with low frequency allocation reuse, the interference sources are very

close. Even a neighboring cell may be an interferer by sharing at least some of the frequencies. In

that kind of situations the C/I is very low when the collisions occur. In order to guarantee an adequate

quality, the collision probability has to be made low. The closer the interferers, the more infrequent

the collisions must be in order to maintain a proper quality. The collision probability depends on the

load of the hopping frequencies called a frequency load. The frequency load describes the probability

that a frequency channel is used for transmission at one cell at one time.

The frequency loadis a product of two other loads: the average busy hour TCH occupancy, which

should in most cases be equal to the hard blocking loadthat is presented in Section 5.4.2, and the

fractional loadthat is presented in Section 5.4.3. The frequency load can be written as

L L Lfreq HW frac= , ( 5.3 )

where:

Lfreq= frequency load

LHW= the busy hour average hard blocking load

Lfrac= fractional load


36/80


Version Date Page

1.0.0 Oct 23, 1998 36/80

Each frequency allocation reuse corresponds to a different C/I at the cell border, thus requiring a

different maximum allowed frequency load in order to keep the collision probability low enough.

5.4.2 Hard Blocking Load

Hard blocking means that all the available traffic channels in the cell are in use and all the new callattempts fail because of the lack of available traffic channels. If it is assumed that the call attempts

occur randomly, then the number of call attempts in a time interval is Poisson distributed. If the call

attempts are Poisson distributed and the length of the calls is exponentially distributed, then the hard

blocking probability (that is also known as the grade of service) can be calculated by using the Erlang

B formula

B

T

N

T

n

N

TCH

n

n

N

TCH

TCH

=

=

!

!0

, ( 5.4 )

where:

B= hard blocking probability

T= offered traffic (Erl)

NTCH= number of TCHs in the cell

In order not to exceed the predefined hard blocking probability, the average busy hour TCH

occupancy may not exceed the threshold defined by the offered traffic at the desired blocking

probability and the number of TCHs. When determining the hard blocking load, only the non-BCCH

TRXs should be considered as illustrated in Figure 5-6. Thats because the BCCH TRX is non-

hopping in RF hopping cell and the calculation of the loads is only relevant in soft blocking limited

network. Currently soft blocking limited BB hopping networks should not be designed because of the

lack of the gatekeeper algorithm, which prohibits the initialisation of new calls if the load in thenetwork is about to exceed the load threshold at the soft blocking limit. The hard blocking load is

calculated as

LT

NHW

hopTCH

hopTCH

= , ( 5.5 )

where:

LHW = hard blocking load

ThopTCH= average number of used TCHs in the busy hour

NhopTCH= total number of TCHs in the hopping TRXs


37/80


Version Date Page

1.0.0 Oct 23, 1998 37/80

BCCH SDCCH SDCCH TCH TCHTCHTCH

TCH TCH TCH TCH TCHTCHTCH



TRX-1

TRX-2

TRX-3

TRX-4

f1

f2,f3,f4

f3,f4,f2

f4,f2,f3

TCH

TCH

TCH

TCH

Active slots Empty slots

75 % 25 %

Load on the BCCH TRXnot considered, sincethe BCCH frequenciesare planned separately

Figure 5-6. Hard blocking load of 75% on RF hopping TRXs.

The average busy hour TCH load, as defined in Equation (5.5), can be used as the maximum TCH

occupancy. In reality, there are times when the TCH occupancy is over the busy hour averageLHW.

However this happens randomly and since the LHWlimit is an average there is about an equal time in

which the load is less than the LHW. If the offered traffic is Poisson distributed, the frequency

allocation can be quite safely dimensioned by using theLHWas the maximum TCH occupancy. In an

environment where the offered traffic is known not to be randomly generated, a higher figure should

be used.

5.4.3 Fractional Load

Fractional loading means that the cell has been allocated more frequencies than there are TRXs as

illustrated in Figure 5-7. This is only possible for RF hopping TRXs. The fractional loading is very

useful when the number of TRXs is low. By utilising fractional loading, it is possible to provide

enough frequencies to hop over (to get FH gain) to even a cell with just one hopping TRX. Fractional

load can be calculated as

L N

Nfrac

TRX

freqs cell

=/

, ( 5.6 )

where:

Lfrac= fractional load

NTRX= number of TRXs in a cell

Nfreqs/cell= number of frequencies allocated to a cell (MA-listlength)


38/80


Version Date Page

1.0.0 Oct 23, 1998 38/80

BCCHTRX-1

TRX-2

TRX-3

TRX-4

f1

f2, f3, f4, f5, f6

f

RF PLANNING .pdf

Documents