C
W
C
Introduction and Motivation
ARL - Modern Concept of Telecommunications
Optimization Problems (PHY)
Nothing Is Ideal (Trade-offs)
Adaptive Modulation Scheme (AMS)
Coding and AMS
Adaptive Multicarrier Modulation and Adaptive CDMA
Other Techniques at PHY
ARL in Today’s Cellular Systems
Conclusions
Outline
N.B.: Adaptive Radio Links ≅ ARL

C
W
C
Introduction and Motivation

The transmission problem
– power and bandwidth constraint
→→→→ channel capacity
– delay and complexity constraint

The noise source
– distortion in time and frequency (noise, fading, multipath … etc.)
→→→→ channel state information (CSI)
– multiuser interference
→→→→ traffic
– time-varying

Solution
– adaptive receiver
– adaptive transmitter
– combination of both

Adaptive transmitter only
– ideally fading channel → Gaussian channel
aposterioryapriory

C
W
C

Conventional solution
– adaptive receiver only, so that the system design for the worth case or average channel
– wasting the channel capacity

New solution
– exploit all the available (time-varying) channel capacity
– adaptive transmitter with partial/perfect channel state information and/or traffic situation
to compensate it apriory
– transmitter = function( channel( time ) , traffic( time ))

CSI is obtained through
– channel reciprocity (TDD) - open loop adaptation
• relatively faster fading channels but interference limited
– channel feedback (FDD) - closed loop adaptation
• feedback is usually limited (latency, overhead)
– cf. power control

Restrictions
– point-to-point duplex connection
Introduction and Motivation - Cont.
Tx RxCh
Tx RxCh
Tx Rx
Rx Tx
BS MS

C
W
C

What is adaptive
– physical layer
• adaptive modulation scheme (AMS) - power and data rate
• coding - both source and channel
• antennas (e.g. adaptive beamforming, or antenna switching)
– link layer
• radio resource management - avoid or minimize collisions, retransmissions,
interference (e.g. Dynamic Frequency Allocation, adaptive MAC, ARQ)
– higher layers
• routing (e.g. Ad-Hoc networks, ODMA)
– adaptive users

Areas
– information theory
– detection theory
– estimation theory
– signal processing
– etc.
ARL - Modern Concept of Telecommunications
Communications ≈30.000/65.000
ARL ≈10.000
MCM ≈2.000
AMS ≈200
bit loading ≈50
IEL Online

C
W
C

Goal
– maximize link throughput - spectral efficiency (bits/s/Hz)
(in fact #of users/link)
– maximize network throughput - area spectral efficiency (bits/s/Hz/m2)
– minimize power (less stringent SNR requirements, or less complex computation)

Problem
– distribution of information in Time-Frequency-Space ?

Data sources (interface to PHY)
– variable bit rate source (VRS), e.g. data transfer
– constant rate source (CRS), e.g. speech
– available rate source (ARS), e.g. video

QoS
– at PHY - delay (maximum, jitter) and BER (average) - e.g. voice ver. data
– higher layers - goodput
– multimedia services - distortion rather than BER
ARL - Modern Concept of Telecommunications - Cont.
time
frequency
space
time
frequency

C
W
C

Assumptions
– fading channel with being the instantaneous SNR with distribution
– variable power with average power constraint
– variable rate [bits/symbol] as to vary symbol period is impractical

Problem definition#1 (VBR)
– maximize the average throughput
– subject to average power constraint
– subject to instantaneous BER constraint
(more restrictive than average BER)

Problem definition#2 (CBR)
– minimize the average power
– subject to average throughput
– subject to average BER constraint

Solution
– method of LaGrange multipliers to obtain optimum rate/power adaptation policy
– leads to a water-filling (in different dimensions)

Practical restriction
– define fading regions - cutoff rate
– average fading rate duration (AFRD) → channel model as a Marcov process
Optimization Problems (PHY)
γ )(γp
)(γS S
)(γk

∞
0
)()( γγγ dpk

=∞
0
)()( SdpS γγγ
BERBER =)(γ
¡
∞
0
)()( γγγ dpS
Kdpk =
¡
∞
0
)()( γγγ
BERdpBER =
¡
∞
0
)()( γγγ
}0{},;{)( /∪<+Ζ∈∈ Niiik γ
),[)
1
,
0
[)
0
,0[ ∞∪∪∪=+ℜ∈ Nγγγγγ 0γ

C
W
C
Nothing Is Ideal (Trade-offs)
Principle of uncertainty (physics) - arise with a practical implementation

General trade-offs (for Communications Engineering)
– BER ver. throughput (implied by channel coding theorem)
– performance ver. delay and complexity (not implied by Shannon theory)
– power efficiency ver. spectral efficiency (e.g. CDMA ver. OFDM)
– spectral efficiency ver. area spectral efficiency

Antennas
– physical size ver. performance

Coding and modulation
– channel coding ver. source coding (not implied by Shannon theory)
– modulation level ver. coding
– coding gain ver. complexity
– amount of feedback ver. complexity
– single link ver. cellular network

System related
– CDMA - spreading ver. coding (CDMA, MC-CDMA)
– OFDM - crest factor ver. out of band radiation
spreading
coding loading
?
hardware
?
software
baseband
passband

C
W
C
Adaptive Modulation Scheme (AMS)
AMS is
– technique to approach the channel capacity by varying (cf. solution to optimization problem)
• power only (cf. power control)
• rate only
• both, power and rate.

Special attention to
– AMS with coding (discussed later)
– AMS in network
• optimization of one link ↑ spectral efficiency of that user, however,
create MUI and ↓ area spectral efficiency (tradeoff)
– AMS with multiple antennas
– AMS as a multicarrier modulation and/or with spreading (discussed later)

Limitations
– Doppler spread
• fast fading - the channel cannot be tracked and the performance is poor
• slow fading - long outage periods infers large data buffers and significant link latency
– delay spread
• complexity ver. performance tradeoff in adaptive multicarrier modulation
• single carrier modulation - problem to explicitly evaluate BER as a function of SNR

C
W
C
Adaptive Modulation Scheme (AMS) - Cont.
System model - both TDD and FDD

Steps to be taken
– estimate the channel
• outdated or erroneous estimates significantly impair the performance
• hence, prediction rather than estimation
– select new transmission format
• instantaneous/average SNR based
• instantaneous/average BER based (decoder)
– signaling of the new format to the receiver (overhead) or blind detection

Effects of AMS
– bursts of errors removed, and constant BER supports well-established codes to be used
– if the delay is not a problem, the gain of AMS can be enormous
– in practice, usually adaptive MQAM (e.g. no Tx, BPSK, 4QAM, 16QAM)
AMS and coding Tx power Channel
channel estimate
Demodulation
and decoding
data data
delay τ
error ε

C
W
C
Coding and AMS
Source coding
– source and channel code trade-off
– e.g. unequal error protection, layered coding for multimedia

Block codes (variable block length)

Convolutional codes (variable interleaving and puncturing)

Coset codes (separation of coding and modulation) (6dB from capacity)
– trellis and lattice codes
(variable coset size)

Turbo codes (3dB from capacity)
– cannot separate coding and modulation
– BER curves necessary - upper bound
– additional constraint of block length
– adapt the channel encoder itself

C
W
C
Adaptive Multicarrier Modulation and Adaptive CDMA
Adaptive MCM
– in fact, MCM means that frequency dimension is available
– by SVD a set of independent parallel flat fading channels
– joint adaptation of subcarriers - bit and power loading
– mitigate the latency problem over slow fading channels
– decomposition also via DFT - requires finite block length
otherwise backward/forward adaptation to attain the channel capacity
– other degrees of freedom
• number of subcarriers (related to coherence bandwidth)
• cyclic prefix

Adaptive CDMA
– degrees of freedom
• multicode
• variable processing gain
• multilevel modulation

Adaptive MCM-CDMA
– great flexibility (the degrees of freedom)
– ranges from OFDM to FH/DS-CDMA

C
W
C
Other Techniques at PHY
Precoding
– pre-Rake, or design of spreading codes to force zero MUI in GMC-CDMA

Predistortion and preequalization
– prevent noise enhancement at the receiver
– power limited channel inversion at the transmitter (keep constant SNR)
– receiver can be simpler

Beamforming and smart antennas
– MIMO → SISO structure
– 1D coding followed by precoding (antenna weights) can achieve the capacity

Uncorrelated antennas
– antenna selection/switching - good performance and complexity tradeoff
– MCM preceding each antenna create the set of flat fading channels

Variable packet length
– related to MAC and also to coding

C
W
C

Today’s cellular systems
– assure 90-95% coverage for certain QoS
– hence, most area excessive SNR - support of higher data rates

Constraints
– adaptation on frame by frame basis
– slow feedback (≈ 10-100 ms)

Historical note
– the first idea late 60’s, however, short lived due to hardware limitations and lack of good
channel estimation techniques
– renewed interested late 80’s for meteor-burst communications
– early 90’s, variable rate MQAM by Steel and Webb
ARL in Today’s Cellular Systems
WCDMA CDMA2000 IS-95B
variable spreading
and coding
variable spreading
and coding
code aggregation
GPRS GPRS-136 EGPRS
adaptive coding and
slots aggregation
adaptive modulation and
slots aggregation
adaptive modulation and
slots aggregation

C
W
C
Conclusions
ARL is a modern concept of communications
– the transmission format is change according to predicted channel conditions
– the traffic is scheduled in order to avoid interference

The main motivation is the lack of spectrum (can be only worse in future)
– the spectrum shall be used as efficiently as possible

Adaptive transmitter to approach time-varying channel capacity
– the more degrees of freedom the better the performance
– at PHY layer
• water-filing in time (adaptive modulation)
• water-filing in frequency (bit and power loading)
• water-filing in time and frequency (not studied, yet)
– upper layers
• avoid interference and optimize routing

ARL principles in all current standards
– cellular
• WCDMA, CDMA2000, IS-95, GPRS, GPRS-136, EGPRS
– broadband
• HIPERLAN/2, IEEE 802.11

Adaptive Radio Links

More Related Content

What's hot (20)

Viewers also liked (10)

Similar to Adaptive Radio Links (20)

More from Pavel Loskot (8)

Recently uploaded (20)

Adaptive Radio Links