1. technology and computing
2. hardware
3. computer networking
4. wireless technology

EE360: Multiuser Wireless Systems and Networks
Lecture 5 Outline
 Announcements
Project proposals due 1/27
 Makeup lecture for 2/10 (previous Friday 2/7, time TBD)
 Reading, Supplemental reading, and class pace

 Small
cells, HetNets, and SoN
 Shannon Capacity of Cellular Systems
 Area Spectral Efficiency
 Multiuser Detection in cellular
 MIMO in Cellular
Review of Last Lecture
 Multiuser

Tradeoff in performance versus complexity
 Multiuser

Detection
OFDM Techniques
OFDMA most common
 Cellular
System Overview
Reuse frequencies to increase spectral efficiency
 Sophisticated PHY layer
 Centralized control

 Standards
4G is LTE: uses OFDMA and MIMO to achieve
 50-100 Mbps
 10-20 MHz of spectral available: hard to compete with WiFi

Rethinking “Cells” in Cellular
Small
Cell
Coop
MIMO
Relay
DAS

How should cellular
systems be designed?
Will gains in practice be
big or incremental; in
capacity or coverage?
Traditional cellular design “interference-limited”






MIMO/multiuser detection removes interference
Cooperating BSs form MIMO array: what is a cell? Next
Lecture
Distributed antennas move BS towards boundary
Relays change cell shape and boundaries
Mobile relaying, virtual MIMO, analog network coding.
Small cells create a cell within a cell
Are small cells the solution to
increase cellular system capacity?
Yes, with reuse one and adaptive techniques
(Alouini/Goldsmith 1999)
Area Spectral Efficiency
A=.25D2p
 S/I increases with reuse distance (increases link capacity).
 Tradeoff between reuse distance and link spectral efficiency (bps/Hz).
 Area Spectral Efficiency: Ae=SRi/(.25D2p) bps/Hz/Km2.
The Future Cellular Network: Hierarchical
Architecture
Today’s architecture
MACRO: solving
initial coverage • 3M Macrocells serving 5 billion users
issue, existing
network
10x Lower HW COST
PICO: solving
street, enterprise
& home
coverage/capacity
issue
FEMTO: solving
enterprise &
home
Picocell
Macrocell
coverage/capacity
issue
Femtocell
10x
CAPACITY
Improvement
Near 100%
COVERAGE
Managing interference
between cells is hard
Deployment Challenges
Deploying One Macrocell
New site verification
Effort
(MD – Man
Day)
1
On site visit: site details verification
0.5
On site visit: RF survey
0.5
New site RF plan
2
Neighbors, frequency,
preamble/scrambling code plan
0.5
Interference analyses on surrounding
sites
0.5
Capacity analyses
0.5
Handover analyses
0.5
Implementation on new node(s)
Field measurements and verification
Optimization
Total activities
0.5
5M Pico base stations in 2015 (ABI)
• 37.5M Man Days = 103k Man Years
•Exorbitant costs
•Where to find so many engineers?
Small cell deployments require
automated self-configuration
via software
2
2
7.5 man days
Basic premise of selforganizing networks (SoN)
SON for LTE small cells
Mobile Gateway
Or Cloud
Node
Installation
Self
Healing
SoN
Server
Initial
Measurements
IP Network
Self
Configuration
Measurement
SON
Server
Self
Optimization
X2
X2
Small cell BS
Macrocell BS
X2
X2
Algorithmic Challenge: Complexity

Optimal channel allocation was NP hard
in 2nd-generation (voice) IS-54 systems

Now we have MIMO, multiple frequency
bands, hierarchical networks, …

But convex optimization has advanced a
lot in the last 20 years
Innovation needed to tame the complexity
Cellular System Capacity

Shannon Capacity



User Capacity




Shannon capacity does no incorporate reuse distance.
Wyner capacity: capacity of a TDMA systems with joint
base station processing
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.
Area Spectral Efficiency

Capacity per unit area
In practice, all techniques have roughly the same capacity for voice, but
flexibility of OFDM/MIMO supports more heterogeneous users
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


TDMA systems with joint base station processing
Multicell Capacity
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.)
Wyner Uplink Capacity

Linear or hexagonal cells

Received signal at base station (N total users)



Propagation for out-of-cell interference captured by a
Average power constraint: E
Capacity CN defined as largest achievable rate (N users)
Linear Array

Theorem:
lim N  CN  C * (a )
for
Optimal scheme uses TDMA within a cell
- Users transmit in 1/K timeslots; power KP
Treats co-channel signals as interference:
Results

Alternate TDMA

CDMA w/ MMSE
Channel Reuse in Cellular Systems
• Channel Reuse in Cellular Systems
• Motivation: power falloff with transmission distance
• Pro: increase system spectral efficiency
• Con: co-channel interference (CCI)
• “Channel”: time slot, frequency band, (semi)-orthogonal code ...
• Cellular Systems with different multiple-access techniques
• CDMA (IS-95, CDMA2000): weak CCI, channel reuse in every cell
• codes designed with a single and narrow autocorrelation peak
• TDMA (GSM), FDMA (AMPS): much stronger CCI
• a minimum reuse distance required to support target SINR
• Channel reuse: traditionally a fixed system design parameter
15
Adaptive Channel Reuse
• Tradeoff
• Large reuse distance reduces CCI
• Small reuse distance increases bandwidth allocation
• Related work
• [Frodigh 92] Propagation model with path-loss only
channel assignment based on sub-cell compatibility
• [Horikawa 05] Adaptive guard interval control
special case of adaptive channel reuse in TDMA systems
• Current work
• Propagation models incorporating time variation of wireless channels
static (AWGN) channel, fast fading and slow fading
• Channel reuse in cooperative cellular systems (network MIMO)
compare with single base station processing
16
System Model
• Linear cellular array, one-dimensional, downlink, single cell
processing
best models the system along a highway [Wyner 1994]
• Full cooperation leads to fundamental performance limit
• More practical scheme: adjacent
17 base station cooperation
Channel Assignment
• Intra-cell FDMA, K users per cell, total bandwidth in the system K·Bm
• Bandwidth allocated to each user
• maxium bandwidth Bm, corresponding to channel reuse in each cell
• may opt for a fraction of bandwidth, based on channel strength
• increased reuse distance, reduced CCI & possibly higher rate
18
Single Base Station Transmission: AWGN
• Path loss only, receive power
Pr (d )  A  Pt  d 
A: path loss at unit distance
γ : path-loss exponent
 ( d , ) 
• Receive SINR
d 
 2L  d    2L  d 
 NAP0t
L: cell radius. N0: noise power
• Optimal reuse factor
• Observations
arg max Bm  log 1   (d , )
• Mobile close to base station -> strong channel, small reuse distance
• Reuse factor changes (1 -> ½)19at transition distance dT = 0.62 mile
Rayleigh Fast Fading Channel
• Environment with rich scatters
• Applies if channel coherence time shorte
than delay constraint
Pr  A  g  Pt  d 
• Receive power
g: exponentially distributed r.v.
• Optimal reuse factor
arg max Bm  E g log 1   (d , , g )
• Lower bound: random signal
Upper bound: random interference
• Observations
• AWGN and fast fading yield similar performance
reuse factor changes (1 -> ½) at transition distance dT = 0.65 mile
• Both “sandwiched” by same upper/lower bounds (small gap in between)
20
Rayleigh Slow Fading Channel
• Stringent delay constraint, entire
codeword falls in one fading state
• Optimal reuse factor
arg max Bm  log 1   (d , , g )
• Compare with AWGN/slow fading:
optimal reuse factor only depends on
distance between mobile and base station
• Observations
• Optimal reuse factor random at each distance, also depends on fading
• Larger reuse distance (1/τ > 2) needed when mobiles close to cell edge
21
Base Station Cooperation: AWGN
• Adjacent base station cooperation, effectively 2×1 MISO system
• Channel gain vectors: signal
 d0 2

h0  
 2 
(2 L  d 0 ) 

h1I, 2
  
interference

  2 L  d  2 
0



 2
2L
   2 L  d 0  

w  w( j )  h( j ) h( j )
• Transmitter beamforming
• optimal for isolated MISO system with per-base power constraint
• suboptimal when interference present
• an initial choice to gain insight into system design
22
Performance Comparison
Observations
• no reuse channel in adjacent cell: to
avoid base station serving user and
interferer at the same time
• reuse factor ½ optimal at all d:
suppressing CCI without overly shrinking
the bandwidth allocation
• bandwidth reduction (1-> ½) overshadows benefit from cooperation
• Advantage of cooperation over single cell transmission: only prominent when users
share the channel; limited with intra-cell TD/FD [Liang 06]
• Remedy: allow more base stations to cooperate
in the extreme case of full cooperation, channel reuse in every cell
23
Area Spectral Efficiency
BASE
STATION
A=.25D2p =



S/I increases with reuse distance.
For BER fixed, tradeoff between reuse distance and link
spectral efficiency (bps/Hz).
Area Spectral Efficiency: Ae=SRi/(.25D2p) bps/Hz/Km2.
ASE with Adaptive Modulation

Users adapt their rates (and powers) relative to
S/I variation.

S/I distribution for each user based on
propagation and interference models.
  S / S
d



d
i
Computed for extreme interference conditions.
Simulated for average interference conditions.
The maximum rate Ri for each user in a cell is
computed from its S/I distribution.

For narrowband system use adaptive MQAM analysis
Propagation Model

Two-slope path loss model:
K
S (d ) 
S,
d (1  d / g )
r
a
b
t

Slow fading model: log-normal shadowing

Fast fading model: Nakagami-m


Models Rayleigh and approximates Ricean.
ASE maximized with reuse distance of one!

Adaptive modulation compensate for interference
Average Area Spectral
Efficiency
[Bps/Hz/Km2]
ASE vs. Cell Radius
fc=2 GHz
1
10
D=4R
D=6R
D=8R
0
10
0.1
0.2
0.3
0.4
0.5
0.6
Cell Radius R [Km]
0.7
0.8
0.9
1
MUD, Smart Antennas
and MIMO in Cellular
MUD in Cellular
In the uplink scenario, the BS RX must
decode all K desired users, while
suppressing other-cell interference from
many independent users. Because it is
challenging to dynamically synchronize
all K desired users, they generally
transmit asynchronously with respect to
each other, making orthogonal
spreading codes unviable.
In the downlink scenario, each RX
only needs to decode its own signal,
while suppressing other-cell
interference from just a few dominant
neighboring cells. Because all K users’
signals originate at the base station,
the link is synchronous and the K – 1
intracell interferers can be
orthogonalized at the base station
transmitter. Typically, though, some
orthogonality is lost in the channel.
MIMO in Cellular:
Performance Benefits

Antenna gain  extended battery life, extended
range, and higher throughput

Diversity gain  improved reliability, more
robust operation of services

Interference suppression (TXBF)  improved
quality, reliability, and robustness

Multiplexing gain  higher data rates

Reduced interference to other systems
Optimal use of MIMO in cellular systems, especially
given practical constraints, remains an open problem
Sectorization and
Smart Antennas
5
2
5
3
5




8C32810.46-Cimini-7/98
7
6
5
1
4
5
5
1200 sectoring reduces interference by one third
Requires base station handoff between sectors
Capacity increase commensurate with shrinking cell size
Smart antennas typically combine sectorization with an
intelligent choice of sectors
Beam Steering
SIGNAL
INTERFERENCE
INTERFERENCE

SIGNAL
OUTPUT
BEAMFORMING
WEIGHTS
Beamforming weights used to place nulls in up
to NR directions


Can also enhance gain in direction of desired signal
Requires AOA information for signal and interferers
Multiplexing/diversity/interference
cancellation tradeoffs
Interference
Stream 2
Stream 1



Spatial multiplexing provides for multiple data streams
TX beamforming and RX diversity provide robustness to
fading
TX beamforming and RX nulling cancel interference

Can also use DSP techniques to remove interference post-detection
Optimal use of antennas in wireless networks unknown
Diversity vs. Interference Cancellation
x1(t)
x2(t)
wt1(t)
wt2(t)
r1(t)
r2(t)
wr1(t)
wr2(t)
sD(t)
+
xM(t)
wtT(t)
Nt transmit antennas
rR(t)
wrR(t)
NR receive antennas
Romero and Goldsmith: Performance comparison of MRC and IC
Under transmit diversity, IEEE Trans. Wireless Comm., May 2009
y(t)
Diversity/IC Tradeoffs

NR antennas at the RX provide NR-fold
diversity gain in fading


Get NTNR diversity gain in MIMO system
Can also be used to null out NR interferers via
beam-steering

Beam steering at TX reduces interference at RX

Antennas can be divided between diversity
combining and interference cancellation

Can determine optimal antenna array
processing to minimize outage probability
Diversity Combining Techniques

MRC diversity achieves maximum SNR in
fading channels.

MRC is suboptimal for maximizing SINR
in channels with fading and interference

Optimal Combining (OC) maximizes
SINR in both fading and interference
 Requires
knowledge of all desired and
interferer channel gains at each antenna
SIR Distribution and Pout

Distribution of  obtained using similar analysis
as MRC based on MGF techniques.

Leads to closed-form expression for Pout.


Similar in form to that for MRC
Fo L>N, OC with equal average interference
powers achieves the same performance as MRC
with N −1 fewer interferers.
Performance Analysis for IC

Assume that N antennas perfectly cancel
N-1 strongest interferers
 General
fading assumed for desired signal
 Rayleigh fading assumed for interferers

Performance impacted by remaining
interferers and noise
 Distribution
of the residual interference
dictated by order statistics
SINR and Outage Probability

The MGF for the interference can be computed
in closed form


pdf is obtained from MGF by differentiation
Can express outage probability in terms of
desired signal and interference as
Pout |Y  y  P( X   ( y   ))  1  e
2

  ( y  2 ) / Ps
Unconditional Pout obtained as
Pout  1  e   ( y 

2
) / Ps
 y / Ps
e
fY ( y)dy

0
Obtain closed-form expressions for most fading distributions
OC vs. MRC for Rician fading
IC vs MRC as function of No. Ints
Diversity/IC Tradeoffs
Distributed Antennas
in Cellular
Distributed Antennas (DAS) in
Cellular

Basic Premise:

Distribute BS antennas throughout cell



Rather than just at the center
Antennas connect to BS through wireless/wireline
links
Performance benefits



DAS
Capacity
Coverage
Power consumption
Average Ergodic Rate

Assume full CSIT at BS of gains for all antenna ports

Downlink is a MIMO broadcast channel with full CSIR

Expected rate is


 N
fi



Ccsit ( P)  Eu Esh log 2 1  S  I 1
a


D
(
p
,
u
)
i









2




Average over user location and shadowing
DAS optimization


Where to place antennas
Goal: maximize ergodic rate
p2
p7
p3
p1
p6
p4
p5
Solve via Stochastic Gradients

Stochastic gradient method to find optimal
placement
1.
2.
3.
4.
5.
Initialize the location of the ports randomly inside the
coverage region and set t=0.
Generate one realization of the shadowing vector f(t)
based on the probabilistic model that we have for
shadowing
Generate a random location u(t), based on the
geographical distribution of the users inside the cell
Update the location vector as Pt 1  Pt   C (u (t ), f (t ), P)
P
Pt
Let t = t +1 and repeat from step 2 until convergence.
Gradient Trajectory

N = 3 (three nodes)

Circular cell size of radius
R = 1000m

Independent log-Normal
shadow fading

Path-loss exponent: a=4

Objective to maximize :
average ergodic rate with
CSIT
Power efficiency gains

Power gain for optimal placement versus central placement

Three antennas
Non-circular layout

For typical path-loss exponents 2<α<6, and for N>5,
optimal antenna deployment layout is not circular
N = 12, α = 5
N = 6, α = 5
Interference Effect

Impact of intercell interference
fi
i 1 D( p , u )a
i
SINR 
6
N
fi
2



 j 1 i 1 j D( p j , u )a
i
N

 j is the interference coefficient from cell j


Autocorrelation of neighboring cell codes for CDMA systems
Set to 1 for LTE(OFDM) systems with frequency reuse of one.
Interference Effect
The optimal layout shrinks towards the center of
the cell as the interference coefficient increases
Power Allocation

Prior results used same fixed power for all nodes

Can jointly optimize power allocation and node placement

Given a sum power constraint on the nodes within a cell, the
primal-dual algorithm solves the joint optimization

For N=7 the optimal layout is the same: one node in the
center and six nodes in a circle around it.

Optimal power of nodes around the central node unchanged
Power Allocation Results
N = 7 nodes

For larger interference and in high path-loss, central node
transmits at much higher power than distributed nodes
Area Spectral Efficiency

Average user rate/unit bandwidth/unit area (bps/Hz/Km2)

Captures effect of cell size on spectral efficiency and interference
• ASE typically increases as
cell size decreases
• Optimal placement leads to
much higher gains as cell size
shrinks vs. random placement
Virtual MIMO and
CoMP in Cellular
Virtual/Network MIMO in Cellular
Many open problems
for next-gen systems
Will gains in practice be
big or incremental; in
capacity or coverage?

Network MIMO: Cooperating BSs form a MIMO array
 Downlink is a MIMO BC, uplink is a MIMO MAC
 Can treat “interference” as known signal (DPC) or noise
 Can cluster cells and cooperate between clusters

Mobiles can cooperate via relaying, virtual MIMO,
conferencing, analog network coding, …
Design Issues: CSI, delay, backhaul, complexity

Open design questions

Single Cluster
Effect of impairments (finite capacity, delay) on the backbone
connecting APs:
 Effects of reduced feedback (imperfect CSI) at the APs.
 Performance improvement from cooperation among mobile
terminals
 Optimal degrees of freedom allocation


Multiple Clusters
How many cells should form a cluster?
How should interference be treated? Cancelled spatially or via
DSP?
 How should MIMO and virtual MIMO be utilized: capacity vs.
diversity vs interference cancellation tradeoffs


Cooperative Multipoint (CoMP)
Part of LTE Standard
- not yet implemented

"Coordinated multipoint: Concepts, performance, and field trial results"
Communications Magazine, IEEE , vol.49, no.2, pp.102-111, February 2011
Summary

HetNets the key to increasing capacity of cellular systems
– require automated self-organization (SoN)

Smart antennas, MIMO, and multiuser detection have a
key role to play in future cellular system design.

Limited results for Shannon capacity of cellular systems

Challenge is how to deal with interference

Area spectral efficiency a good metric for capturing impact
of small cells and frequency reuse

Distributed antennas (DAS) leads to large performance
gains, CoMP not so promising.