05 RN31545EN30GLA0 Capacity Dimensioning

1 Nokia Siemens Networks CN31545EN30GLA0

Course Content

WCDMA & HSPA fundamentals

Radio network planning fundamentals

Radio network planning process

Coverage dimensioning

Capacity dimensioning

Coverage & capacity planning

Coverage & capacity improvements

NSN radio network solution

Site Solutions & Site Planning

Initial Parameter Planning


Module Objectives

At the end of the module you will be able to:

Understand basic traffic modeling

Calculate air interface capacity & load


Air Interface Capacity Dimensioning

Traffic estimate & model

Air interface dimensioning

DCH load calculation

HSDPA capacity

HSUPA capacity

Basic Traffic Model

Air Interface

Dimensioning

Channel Card

Dimensioning

RNC

Dimensioning

Iub

Dimensioning

Iu

Dimensioning

Iur

Dimensioning

+

Topology Subscribers

Rad

io n

etw

ork

A

cce

ss

netw

ork

Note:

- This Learning Element contains the Air Interface

dimensioning

- The dimensioning of Channel Elements (CE)

can be found in the proceeding Learning Element

- Iux & RNC dimensioning can be found in

RN3003 3G IP Transmission Planning & similar courses


Traffic estimation

The traffic estimation requires information related to the network topology, subscribers & traffic:

Cell Area from Coverage Dimensioning

Subscriber density from Marketing

Subscriber traffic profile from Marketing

Basic Traffic Model

Air Interface

Dimensioning

Channel Card

Dimensioning

RNC

Dimensioning

Iub

Dimensioning

Iu

Dimensioning

Iur

Dimensioning

+


Subs density Cell area Traffic / subscriber

Traffic / cell

Traffic / site


Subscriber density

Operator subscriber density depends on:

Population density

Mobile phone penetration

Operator market share

The subscriber density can be considered quite stable in mature markets

Mobile phone penetration close to 100% for basic services

Major changes possible only when new operators come to the market or with aggressive marketing campaigns

In developing markets fast changes in mobile phone penetration and operator market share


Traffic information

The subscriber density & Subscriber traffic profile are the main requirements for capacity dimensioning

Traffic forecast should be done by analysing the offered Busy Hour traffic per subscriber for different services in each rollout phase

Traffic data:

Voice : Erlang per subscriber during busy hour of the network

Codec bit rate, Voice activity

Video call : Erlang per subscriber during busy hour of the network

Service bit rates

NRT data : Average throughput (kbps) per subscriber during busy hour of the network

Target bit rates


Example: Subscriber traffic profile / traffic estimation

Subscriber traffic profile - Marketing Forecast (Example)

(Average) traffic demand per subscriber in busy hour: Speech telephony: 20 25 mErl Video telephony: 2.5 3.0 mErl SMS 0.3

Data services ~ 600 1000 bps (DL), ~ 75 - 100 bps (UL)

Traffic Estimation (Example)

Coverage Area (Site): 10 km2

Planning Area: 100 km2 & 10 000 subscribers 100 subs/km2 1000 subs/Site

User profile Speech traffic: 25 mErl/subs/BH

NRT data traffic: DL 750 bps/subs/BH, UL 75 bps/subs/BH

Site traffic: Speech - 25 Erl/cell/BH +

NRT data DL - 750 kbps/cell/BH,

NRT data UL - 75 kbps/cell/BH


Traffic model: Erlang B

Traffic model is used to derive the required capacity from average traffic & service quality requirement

RT traffic (speech, video call, video streaming) is commonly modelled with Erlang-B model

Average traffic (Erlangs) A

Blocking probability (%) B

required No. of traffic channels N

NRT traffic (web, email services) can

be modelled as average traffic with

defined overhead

N = number of

Trunks

A Traffic

carried

Traffic

Lost

B


Erlang-B model

Erlang-B model is used for a system without queuing

Assumes random call arrival

The Blocking probability B can be calculated as

A = traffic in Erl

N = required number of traffic channels

1% 2% 3% 4% 5% 6% 7% 8% 9% 10%

5 11 10 10 9 9 9 9 8 8 8

6 13 12 11 11 10 10 10 9 9 9

7 14 13 12 12 11 11 11 10 10 10

8 15 14 14 13 13 12 12 12 11 11

9 17 15 15 14 14 13 13 13 12 12

10 18 17 16 15 15 14 14 14 13 13

11 19 18 17 16 16 15 15 15 14 14

12 20 19 18 18 17 17 16 16 15 15

13 22 20 19 19 18 18 17 17 16 16

14 23 21 21 20 19 19 18 18 17 17

15 24 23 22 21 20 20 19 19 18 18

16 25 24 23 22 21 21 20 20 19 19

17 27 25 24 23 22 22 21 21 20 20

18 28 26 25 24 23 23 22 22 21 21

19 29 27 26 25 24 24 23 23 22 22

20 30 28 27 26 26 25 24 24 23 23

21 31 29 28 27 27 26 25 25 24 24

22 32 31 29 28 28 27 26 26 25 25

23 34 32 30 29 29 28 27 27 26 26

24 35 33 32 31 30 29 28 28 27 27

25 36 34 33 32 31 30 29 29 28 28

26 37 35 34 33 32 31 30 30 29 29

27 38 36 35 34 33 32 31 31 30 29

28 39 37 36 35 34 33 32 32 31 30

29 40 38 37 36 35 34 33 33 32 31

30 42 39 38 37 36 35 34 34 33 32

31 43 41 39 38 37 36 35 35 34 33

32 44 42 40 39 38 37 36 35 35 34

33 45 43 41 40 39 38 37 36 36 35

34 46 44 42 41 40 39 38 37 37 36

35 47 45 43 42 41 40 39 38 38 37

36 48 46 44 43 42 41 40 39 39 38

37 49 47 45 44 43 42 41 40 40 39

38 51 48 46 45 44 43 42 41 40 40

39 52 49 47 46 45 44 43 42 41 41

40 53 50 48 47 46 45 44 43 42 42

41 54 51 50 48 47 46 45 44 43 43

42 55 52 51 49 48 47 46 45 44 43

43 56 53 52 50 49 48 47 46 45 44

44 57 55 53 51 50 49 48 47 46 45

45 58 56 54 52 51 50 49 48 47 46

46 59 57 55 53 52 51 50 49 48 47

47 61 58 56 54 53 52 51 50 49 48

48 62 59 57 55 54 53 52 51 50 49

49 63 60 58 56 55 54 53 52 51 50

50 64 61 59 57 56 55 54 53 52 51

N = required No. of trunks

B = Blocking Probability

N

i

i

N

i

A

N

A

ANB

0 !

!),(

A = Average traffic [Erl]


Packet data modelling

Packet data traffic is a sum of multiple services with different traffic profiles and service quality requirements

Accurate modelling of packet data traffic requires multiple assumptions and complex simulations

Practical packet data traffic model utilises average bit rate with fixed overhead for protocol and QoS

The overhead can assumed to be 27%

This figure includes the L2 re-transmission overhead of 10% and 15% of buffer headroom to avoid overflow (peak to average load ratio headroom) (1+0.10) x

(1+0.15) = 1.265 26.5% overhead

Required bit rate = (1 + Overhead) * Average bit rate


Example: Traffic models

Cell traffic: 25 Erl/cell/BH, 750 kbps/cell/BH

Speech: 25 Erl & 2% blocking 34 traffic channels

NRT data DL: 750 kbps * (1 + 26%) = 945 kbps

NRT data UL: 75 kbps * (1 + 26%) = 94.5 kbps

assumed overhead

for protocol & QoS

10% L2 re-transmission overhead

15% buffer headroom to avoid overflow

(1+0.10) x (1+0.15) = 1.265 26.5% overhead






HSDPA capacity

HSUPA capacity

Basic Traffic Model

Air Interface

Dimensioning

Channel Card

Dimensioning

RNC

Dimensioning

Iub

Dimensioning

Iu

Dimensioning

Iur

Dimensioning

+


Rad

io n

etw

ork

A

cce

ss

netw

ork


Cell load calculation is needed in order to estimate the level of air interface load in the cell

Air interface load depends on service mix, radio propagation conditions, network topology and number of active connections as well as traffic inputs

or load estimation

Service type Bitrate, Eb/N0

Propagation conditions Eb/N0, Orthogonality

Network topology Little i (other cells Interference / own cell Interference

Air interface load Link budget

Cell range

Load/cell Load estimation Traffic inputs

Load Calculation Introduction


Air interface capacity

WCDMA air interface capacity can be estimated with system simulations and/or analytical load calculations

System simulations provide a complete system model and possibility to model system specific parameters and network layout

Complex tools, not feasible to use for dimensioning

Dimensioning can be done with pre-analysed results Limited possibility to change system parameters

Analytical models utilise system and environment specific input parameters and simple models

Simple analysis can be done as part of dimensioning process

Parameters configurable flexible model

Results rely on realistic input parameter values


Load Calculation: Uplink Load

jjbj

j

NE

RWL

1

/

/1

1

0

N

j

jUL L0

j: Activity factor; for Speech some 67% due to VAD/DTX; for Data: 1

Load Lj of subscriber

with Service j

UL

total

Cell Load

Activity Factor

Processing Gain

0

2

4

6

8

10

12

14

16

18

10

20

30

40

50

60

70

80

90

95

98

loading/%

loss/d

BIn

terf

ere

nce M

arg

in [dB

]

UL = 30 50 %

Cell Load [%]

Load Calculation Formulas in analogy to

H. Holma WCDMA for UMTS


Inter-Cell Interference: Little i

In the real environment we will never have separated cell. Therefore, in the load factor calculation the other cell interferences should be taken into account.

This can be introduced by means of the Little i value, which describes how much two cells overlap (bigger overlapping more inter-cell interferences)

Iother

ceinterferen cellown

ceinterferen cellother i

Inter-Cell Interference Ratio

Little i

j

jjb

jj

jUL

NE

RWiLi

1

/

/1

1)1()1(

0


Uplink Load calculation

Simplified UL load equation UL DCH capacity

for 1 service type j only

W/Rj >> (Eb/No)j

Nj: No. of Trunks

Nj x Rj = Cell Throughput = Capacity [kbps]

j

jb

jjULRW

NoENi

/

)/()1(


Downlink Load calculation

The DL capacity can be calculated in a similar manner as the UL capacity from the DL Load

The equations are similar to those of the UL, except two modifications:

Soft Handover Overhead SHO_OH: an Overhead has to be integrated into the calculation due to Soft Handover; in this case two Node Bs require capacity to serve a single user

Orthogonality Factor : In the DL, the Intra-Cell Interference should be theoretically Zero ( Orthogonality of Channelisation Codes);

due to a loss of Orthogonality caused by Multipath transmission,

the Orthogonality Factor has to be taken into account; j = [0 .. 1.0] propagation channel conditions

The DL orthogonality & i are different for each user and average values have to be used in DL load calculations

j

jjb

jjUL

NE

RWiOHSHO

1

/

/1

1)1()_1(

0

Cell Type

Macro Cell 0.4 0.9

Micro Cell > 0.9

typically 50 75 %

No. of Trunks Nj &

Cell Throughput Nj x Rj [kbps]


Little i & SHO overhead

The level of interference received from neighbouring cells depends strongly on

Network layout (site locations, antenna directions & sectorisation)

Propagation environment (propagation slope)

SHO overhead is related to the cell coverage overlap & other cell interference

level

Sectorization HBW SHO

Overhead i = Iother/Iown

1-sector omni 23% 58%

3-sector 90 34% 88%

3-sector 65 27% 66%

3-sector 33 26% 70%

4-sector 90 42% 109%

4-sector 65 31% 76%

4-sector 33 33% 86%

6-sector 90 53% 146%

6-sector 65 42% 105%

6-sector 33 32% 90%

HBW: Half Beam-Width

Interference received from neighbouring cells

simulated DL values


Load Calculation Examples

Load factor for different services has to be calculated separately, total load is then the sum of different services in the cell area

UL/DL single connection load examples are shown in the table below

For example 50 % UL load means on average 50 speech users or about 9 64 kbits/s users/cell in a 3-sector (1+1+1) configuration

Services UL Fractional Load DL Fractional Load

12.2 kbit/s 0,97% 1,00%

64 kbits/s 4,80% 6,21%

128 kbits/s 8,56% 11,07%

384 kbits/s 22,89% 29,59%

Total Load 37,22% 47,87%


Total WBTS DL power R99 traffic

Total DL base station transmit power can be a limiting factor in highly loaded cell

DL

CCCHN

j

jSERVj

j

jb

N

DL

TOT

DL

PL

RW

NEPP

11

1

1

,

0

where,

Lserv is the pathloss of user j. The pathloss is defined as total loss from BTS

transmitter to the receiver

PCCCH is the total common control channel power


Example - Total DL power & load

Total DL power increases exponentially towards 100% of load

Common control channels CCH consumes larger part of DL power

4 W CCCH & 50% load Total power 10.5 W

8 W CCCH & 50% load Total power 18.5 W

PtxTotal with different common channel power

4.0 4.3 4.75.0 5.4

5.9 6.47.0 7.7

8.59.4

10.511.8

13.415.4

17.9

21.3

26.0

33.1

8.0 8.59.1 9.7

10.311.111.9

12.914.0

15.316.7

18.5

20.6

23.1

26.3

30.4

35.9

0.0

5.0

10.0

15.0

20.0

25.0

30.0

35.0

40.0

0%

5%

9%

14%

18%

23%

27%

32%

36%

41%

45%

50%

54%

59%

64%

68%

73%

77%

82%

86%

91%

Downlink DCH load

Ptx

To

tal

4 W

8 W


Example: Load calculation

Is it possible to transmit 34 speech channels in one cell simultaneously with 945 kbps NRT DL data

and 94.5 kbps NRT UL data?

Speech: 34 traffic channels

NRT data: DL = 945 kbps, UL = 94.5 kbps

Fractional load of 12.2 AMR speech:

DL Load = 34 * 1.0% = 34%,

UL load = 34 * 0.97% = 33 %

Fractional load of NRT data (NRT 128 kbps):

DL Load = 750 kbps/128 kbps * 11.07% = 64.9 %,

UL Load = 75 kbps/128 kbps * 8.56% = 5.0 %

total DL load = 97.9%

total UL load = 38%

DL overload!


Example: Capacity analysis

How much DL traffic (in kbps) is possible for a max. allowed DL load of 74%

simultaneously with 25 speech calls ?

Speech traffic of 25 Erlangs corresponds average of 25 calls in the cell

Average speech load: UL = 24%, DL = 25%

Max. cell power 20 W with 2 W pilot allows max. DL load of 74% in the example cell

In average 49% load margin available for NRT data in DL

49% / 11.07% * 128 kbps = 566 kbps

In average 566 kbps DL available for NRT data






HSDPA capacity

HSUPA capacity

Basic Traffic Model

Air Interface

Dimensioning

Channel Card

Dimensioning

RNC

Dimensioning

Iub

Dimensioning

Iu

Dimensioning

Iur

Dimensioning

+


Rad

io n

etw

ork

A

cce

ss

netw

ork


HSDPA dimensioning can be done based on:

Requirement to achieve min. HSDPA cell edge throughput Determined from link budget analysis, SINR at cell edge

Requirement to achieve average HSDPA throughput across the cell Determined by SINR distribution analysis

HSDPA capacity depends on: Available power for HSDPA

Channel conditions

Cell range (pathloss)

Interference level over cell area

HSDPA features & configuration

SINR: Key measure for

HSDPA Peak Data Rate /

Throughput

HSDPA Capacity Introduction / SINR

Geometry Factor

Total Transmit Power

Spreading Factor

Orthogonality factor

Transmitted HS-PDSCH

power

GP

PSFSINR

tot

PDSCHHS

11

16

Geometry Factor G = own Cell Interference / (other Cell Interference + Noise)

SINR: Signal-to-Interference+Noise Ratio


SINR & HSDPA Throughput

The single-user HSDPA throughput versus its average HS-DSCH SINR is

plotted.

Notice that these results include the effect of fast fading & dynamic HS-

DSCH link adaptation (and HARQ).

An average HS-DSCH SINR of 23 dB is required to achieve the maximum data

rate of 3.6 Mbps with 5 HS-PDSCH

codes

Benefit from using more codes (10/15) is only experienced for higher SINR

values >10 dB A

vera

ge s

ingle

-use

r th

roughput [M

bps]

Average SINR (1 HS-PDSCH) [dB]

0.5

1.0

1.5

2.0

2.5

-10 -5 50 10 15 20 25 300

3.0

3.5

4.0

HS-DSCH POWER 7W (OF 15W), 5 CODES, 1RX-1TX, 6MS/1DB LA DELAY/ERROR

Rake, Ped-A, 3km/h

Rake, Veh-A, 3km/h

Rake, Ped-B, 3km/h

MMSE, Ped-A, 3km/h

MMSE, Ped-B, 3km/h

Rake, Veh-A, 30km/h

Average HS-DSCH SINR [dB] Average SINR [dB]

Common cell

edge condition

Inside

macro

cell

Micro cell, LOS,

low interference

Cell

Thro

ughput [M

bps]


HSDPA throughput Orthogonality

Close to the BTS the own cell interference dominates (1/G i 0.9 can be achieved:

in isolated environment

Micro- / Pico- / Femto- Cells

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 1000 2000 3000 4000 5000 6000 7000 8000 9000

Throughput, kbps

Ort

ho

go

nality

10% BTS pow er for HSDPA 50% BTS pow er for HSDPA

80% BTS pow er for HSDPA

116

tot

PDSCHHS

P

PSFSINR

for 1/G i


HSDPA Capacity: HSDPA power

Dynamic Resource Allocation feature: BTS can allocate all unused DL power to HSDPA

All the power available after DCH traffic, HSUPA control channels & common channels can be used for HSDPA

HSDPA power is shared dynamically between HS-SCCH & HS-PDSCH

Time

Power

PtxHSDPA = PWBTS_max PccH_tx - PDCH PHS-PDSCHs = PtxHSDPA PHS-SCCH


HSDPA Capacity G-Factor

The G Factor reflects the distance between the MS & BS antenna thus setting a value for G factor means making assumptions on user location.

A typical range is from -5dB (Cell Edge) to 20dB

Typical G factor distributions (CDF) coming from NSN simulation tools as well as operator field experience are represented in the following chart:

GP

PSFSINR

tot

PDSCHHS

11

16

-20 -10 0

G-factor [dB]

Cum

ula

tive

dis

trib

utio

n f

un

ction

[%

]

10 20 30 400

10

20

30

40

50

60

70

80

90

100

Macrocell(Wallu)Veh-A/Ped-A

Macrocell(Vodafone)

Veh-A/Ped-A

Microcell

(Vodafone)Ped-A

noiseother

own

PI

IG


Cell size & HSDPA cell throughput

Cell size has an effect on HSDPA cell throughput when cell edge pathloss is high (large cell or indoor users)

Increase of BTS power has only limited effect on cell throughput

0

200

400

600

800

1000

1200

1400

100 105 110 115 120 125 130 135 140 145 150 155 160

Cell edge pathloss, dB

HS

DP

A c

ell t

hro

ug

hp

ut

DCH load 10%&20W

DCH load 30%&20W

DCH load 50%&20W

DCH load 10%&40W

DCH load 30%&40W

DCH load 50%&40W

5 codes


HSDPA capacity & Code Multiplexing

HSDPA capacity is influenced by the capabilities of the network and the UE

Number of codes (5, 10, 15) Higher peak bit rate in good conditions Higher cell throughput

Code multiplexing: multiple 5 code UEs can utilise up to 15 codes Higher spectrum efficiency

1.2 Mbps

1.7 Mbps

1.8 Mbps

2.0 Mbps

2.2 Mbps

5 Codes

Cell capability

10 Codes 15 Codes

no code-multiplexing (10/15 code UEs)

code-multiplexing (5 code UEs)


HSDPA Capacity: RU20 features

RU20 features improving the

HSDPA capacity:

64QAM

2x2 MIMO

DC-HSDPA

CS Voice over HSPA

64QAM

max. Peak Rate = 21 Mbps

good channel conditions required to take benefit of 64QAM CQI 26 !

64QAM requires 6 dB higher SNR than 16QAM

average CQI typically 20 in the commercial networks

DC-HSDPA: 1) Improved Load Balancing

2) Frequency Selectivity

3) Reduction of Latency

4) Higher Peak Data Rates

5) Improved Cell Edge User Experienced

10 MHz 1 UE, using 2 RF

Channels:

Peak Rate =

2 x 21 Mbps =

42 Mbps

F1 F2

5 MHz 5 MHz



2x2 MIMO:

Single- or Dual-stream Operation

max. Peak Rate = 28 Mbps

Legacy HSDPA

Cell edge: low SINR

High SINR

Single-stream MIMO

Dual-stream MIMO

Mean C

ell

Thro

ughput [M

bps]

UE

Thro

ughput (P

F)

[kbps]

SISO: Single Input Single Output

RxDiv: Receive Diversity: 1 Tx-, 2 Rx- Antenna(s)

CLM1 2x2: Closed Loop Mode; Single-Stream with Rx- & Tx-Diversity

MIMO 2x2: Dual-Stream MIMO using Spatial Multiplexing

RR: Round Robin

PF: Proportional Fair

PF-RAD-DS: PF scheduling extended by Required

Activity Detection with Delay Sensitivity



Voice over HSPA

[REF. WCDMA for UMTS HSPA Evolution and LTE, HH AT]

Assumed IP Header

Compression

for Voice, SRB

& other services






HSDPA capacity

HSUPA capacity

Basic Traffic Model

Air Interface

Dimensioning

Channel Card

Dimensioning

RNC

Dimensioning

Iub

Dimensioning

Iu

Dimensioning

Iur

Dimensioning

+


Rad

io n

etw

ork

A

cce

ss

netw

ork


HSUPA Capacity HSUPA Cell Throughput

Principle: Example ( Diagram)

max. Load for HSUPA higher than for Rel. 99 DCH* UL(HSUPA) = 80%

1) UL load is shared between HSUPA & R99 DCH users Rel. 99: 50% Load

HSUPA: 80% - 50% = 30% Load

2) UEs distribution inside the cell has impacts on possible C/I;

impacts on cell throughput

here: each UE is allocated an equal share of UL Load LHSUPA_UE = 30% / 5 UE = 6%

* due to Fast Packet Scheduling

LHSUPA_UE: Load per UE

0

2

4

6

8

10

12

0 20 40 60 80 100

Uplink Load (%)

Incre

ase in Inte

rfere

nce (

dB

)

Example Target

Uplink Load

UL Load generated by

R99 DCH

UL Load available

for HSUPA UE


i

IC

Nj

j

jj

UL

1

)/(

11

1

1

HSUPA Capacity HSUPA Cell Throughput

3) UL load is translated to UL C/I

using the UL load equation

C/I: Chip-Energy/Interference = Eb/No Processing Gain* Example: i = 0.65; j(data) = 1

LHSUPA_UE = 6% = (1+ i) / ( 1 + 1 / C/I)

C/I = 1/((1+i)/LHSUPA_UE -1) = 0.051 = - 12.9 dB

4) C/I is translated to HSUPA bit rate

using the Eb/No look-up table

derived from link level simulations

* both in dB; decimal: (Eb/No) / (W/R)

Layer 1

Bit Rate

TTI

(ms)

Physical

Channel

Eb/No with

RxDiv W/R C/I

1920.0 10 2*SF2 0.5 dB 3 dB -2.5 dB

1440.0 10 2*SF2 0.1 dB 4.26 dB -4.16 dB

1024.0 10 2*SF2 0.2 dB 5.74 dB -5.54 dB

512.0 10 2*SF4 0.6 dB 8.75 dB -8.16 dB

384.0 10 1*SF4 0.9 dB 10 dB -9.1 dB

256.0 10 1*SF4 1.1 dB 11.76 dB -10.66 dB

128.0 10 1*SF8 1.9 dB 14.77 dB -12.87 dB

30% HSUPA Load 5 x 128 kbps total


HSUPA Capacity: Example

HSUPA average cell throughput vs. Rel. 99 DCH load

HS

UP

A c

ell

thro

ughput [k

bp

s]

Example:

HSUPA Load = 30%

HSUPA throughput = 5 x 128 kbps



Summary

The Air Interface Capacity dimensioning includes aspects:

Traffic estimation & modelling

Air Interface Load estimation

Rel. 99 / HSDPA / HSUPA Capacity

for each carrier (shared Rel. 99/HSPA or dedicated HSPA)

Capacity strongly depends on:

Interference: Inter-Cell Interference i, SINR

Orthogonality factor

Quality Requirements Eb/No

Power (total Power / HSPA Power)

05 RN31545EN30GLA0 Capacity Dimensioning

Documents

subscribers traffic

basic traffic modeling

offered busy hour traffic

merlsubsbh nrt data

rollout phase traffic

network topology

air interface capacity

network codec