Transcript Wireless Communications Research Overview

EE360: Lecture 15 Outline
Cellular System Capacity

What is capacity?

Defining capacity

Shannon capacity of cellular systems

“Information capacity of symmetric cellular multiple
access channels” Hrishikesh Mandyam

Multicell capacity

User capacity

Summary
What is capacity?

Shannon capacity




Other theoretical capacity definitions



Maximum achievable rate or set of rates with arbitrarily small
probability of error
Coding scheme not achievable, and complexity/delay are
infinite
Tractable formulas exist for point-to-point links, MAC and
degraded broadcast channels, and ad-hoc networks under
various assumptions
Outage capacity
Computation cutoff rate (not very useful)
Capacity definitions used in practice


Achievable rates under practical system assumptions
User capacity under practical system assumptions
Defining Cellular Capacity

Shannon-theoretic definition








Multiuser channels typically assume user coordination and joint
encoding/decoding strategies
Can an optimal coding strategy be found, or should one be
assumed (i.e. TD,FD, or CD)?
What base station(s) should users talk to?
What assumptions should be made about base station
coordination?
Should frequency reuse be fixed or optimized?
Is capacity defined by uplink or downlink?
Capacity becomes very dependent on propagation model
Practical capacity definitions (rates or users)



Typically assume a fixed set of system parameters
Assumptions differ for different systems: comparison hard
Does not provide a performance upper bound
Approaches to Date

Shannon Capacity


Multicell Capacity




TDMA systems with joint base station processing
(Hrish)
Rate region per unit area per cell
Achievable rates determined via Shannon-theoretic
analysis or for practical schemes/constraints
Area spectral efficiency is sum of rates per cell
User Capacity
Calculates how many users can be supported for a given
performance specification.
 Results highly dependent on traffic, voice activity, and
propagation models.
 Can be improved through interference reduction
techniques. (Gilhousen et. al.)

User Capacity

Maximum number of users a cellular
system can support in any cell.

Can be defined for any system.

Typically assumes symmetric data rates,
cells, propagation, and mobility.

Depends on the user specifications and
radio design
 data
rate, BER, modulation, coding, etc.
Multicell Capacity

Multiuser rate region per Hertz divided by
coverage area given reuse distance
( R1 ,..., Rn ) / B
C multicell 
 (.5 RD ) 2

Rate region (R1,…,RN) can be obtained via
Shannon analysis



How to treat interference from other cells
Alternatively, can compute under practical
system assumptions
ASE sums rates in each cell: A   Ri /( B (.5RD )2 )
i
Which link dictates capacity?

Reverse link (MAC)




Forward link (Broadcast)



Noncoherent reception
Independent fading of all users
Requires power control
Coherent demodulation using pilot carrier.
Synchronous combining of multipath.
Conclusion: reverse link has lower capacity



Thus, reverse link dictates capacity
Other cell interference will tend to equalize
performance in each direction.
In asymmetric traffic, forward link will be bottleneck
CDMA User Capacity

Single-Cell System

Similar to MAC user capacity
Eb / N 0 


-
8C32810.44-Cimini-7/98
W /R
W /R
h
 N  1

Eb / N 0 S
( N  1)  (h / S )
G=W/R is processing gain (W is bandwidth, R is data
rate)
h is interference plus noise (assumed fixed)
Assumes power control
Performance improvement through sectorization and
voice activity
Sectorization

Base station omni antenna is divided into M sectors.

Users in other sectors do not cause interference.

Number of users per sector is Ns=N/M (reduces
interference by M).

Requires handoff between sectors at the base station
Voice Activity

Suppress signal when voice user not active.

Voice activity a=.35-.4 (reduces
interference by 60-65%).

Resynchronization for every talk spurt.
 Higher
probability of dropping users.
New Capacity (per cell)
Eb / N 0 
W /R
( N s  1)a  (h / S )
M  W /R  h

 
 N  MN s  M 
a  Eb / N 0  S

Capacity increased proportional to the number of
sectors and inversely proportional to the voice activity
(M/a typically around 8).

Claim: CDMA is competitive with TD for a single-cell

Does not include impact of sectorization on out-of-cell
interference.
Multicell System

Codes reused in every cell.

No power control in forward link
 Interference from adjacent cells can be very strong.

Power control in reverse link
 All users within a cell have same received signal
strength
 Interference from other cells have variable power



Fast fading (interference and signal) neglected (S/I
statistics).
The interferer’s transmit power depends on distance to his
base station.
Received power at desired base depends on distance to
base, propagation, and the interferer’s transmit power.
Reverse Link Interference

Total path loss: propagation (d-4 falloff) and log-normal
shadowing (x is Gaussian, 8 dB STD)
L  10(x /10 ) r 4

Instantaneous interference power
4
I (r0 , rm )  rm  (x 0 x m ) /10
   10
1
S
 r0 

rm is distance to interferer’s base, r0 is distance to desired base

xm is shadowing to interferer base, x0 is shadowing to desired base

S is received power with power control

Power less than 1 since otherwise would handoff to desired base
Average interference power
4


 rm  (x0 x m ) /10
I
  g   10
1 (rm / r0 ) 410(x0 x m ) /10  1 rdA
S
 r0 


-
-
A is the cell area.
r is the user density (r=2Ns/Sqrt[3])
g is voice activity term (equals 1 w.p. a, 0 w.p. 1-a)
Must be integrated against distribution of m, r0, rm, x0, xm
- Simplify distribution of m by assuming minimum
distance.
- r0, rm uniformly distributed.
Claim: I Gaussian since it’s a functional of a 2D white
random process
Mean and Variance

Numerical integration leads to E(I/S)=.247Ns

Second Moment:



Assumes autocorrelation of shadowing is a delta
function and STD is 8 dB.
Numerical integration leads to Var(I/S)=.078Ns
Total interference distribution
Eb / N 0 
W /R
N s 1

i 1
i
 ( I / S )  (h / S )
I Gaussian, i binomial r.v. with probability a
Capacity Calculation

Calculate probability Eb/N0 below target (BER
exceeds target) based on Ns and these statistics.
.

Ns


3
P( BER  10 )  P   i  I / S  d 
 i 1

W /R h
d

Eb / N 0 S
Compute outage probability as a function of Ns.


Assumes target Eb/N0 =5 d=30
Results indicate 60 users/sector with 1% outage
An Alternate Approach

Simulation approach

Includes three rings of interfering cells

Capacity for TDMA and CDMA
compared
 Similar
assumptions about voice activity and
sectorization
 TDMA assumes FH with dynamic channel
allocation
 Results indicate CD greatly outperforms TD

Not surprising given the authors
Capacity degradation

Voice activity changed from .375 to .5, -30%
change

Path-loss changed from 4 to 3, -20% change

Multipath fading added, -45% change

Handoff margin changed from 0 to 6 dB, -40%
change

Power control error changed from 0 to 1 dB, -35%
change
Summary

Multiple definitions of cellular system capacity

True Shannon capacity unknown

Shannon capacity under given system assumptions still
complicated

Using multiccell capacity formulation allows
comparison of apples to apples

User capacity calculations highly dependent on
system assumptions

Easy to skew results in a given direction