QAM 128.rar_64qam matlab_DA9_qam

Paper

PHY Abstraction Methods

for OFDM and NOFDM Systems

Adrian Kliks, Andreas Zalonis, Ioannis Dagres, Andreas Poly d oros, and Hanna Bogucka

Abstract— In the paper various PHY abstraction methods for

both orthogonal and non- orthogonal systems are p resented,

which allow to predict the coded block error rate (BLER)

across the subcarriers transmitting this FEC-coded block for

any given channel realization. First the efficiency of the se-

tems, PHY abstraction methods.

1. Introduction

In the past where multi-modal operation was not an optio n,

the role of performance evaluation (an alytically or by sim-

ulation) was to simply check whether a given signal design

met the pre-specified performance requirements. The av-

erage performance of a system was quantified by using

the topology and the channel macro-characteristics in order

to compute a geometric (or average) signal-to-interference

plus noise ratio (SINR) distribution across the cell. If there

tivation thus arose to have simple analytic forms for these

performance results which would make them amenable to

easy pa rametric optimization .

Once the de sign aspect advances to become multi-modal

and multi-parametric at both sides of the transmission link

(e.g., current orthogonal fre quency division multiplexing

(OFDM) based systems: 3rd generation partnership project-

long term evolution (3GPP-LTE), worldwide interope rabil-

ity for microwave access (WiMAX)), the task of link-

performance evaluation becomes not only germane to the

are currently being developed for various systems in the re-

spective standard iz a tion bodies (e.g., I EEE 802.16m Task

Group [1]). The goal of this type of physical-layer (PHY)

abstraction is to determine the performance of a given

link an d thus avoid the need for extensive simulation.

This “simulation-shortcut” accelera tes the corresponding

system- leve l simulations where a large number of physical-

layer-related links need to be taken into account. Th e ab-

straction should be accurate, compu ta tionally simple, rel-

atively independent of chan nel models, and extensible to

multiplexing (NOFDM) signals, the impu lses used at the

transmitter overlap each other both in time and in frequency

domain, thus they are not orthogonal. The shape and the

signaling time of the applied impulses can be chosen with-

out any r estrictions besides ensuring that the pulses used at

the receiver are biorthogonal to pulses used on the transmit-

ter side. The (NOFDM) systems are the part of the larger

set of generalized mutlica rrier (GMC) systems, where all

of the transmit p arameters can be in general chosen without

any specified restriction. Th us a GMC signal set includes

presented and the proposals for modification of some ab-

straction methods for (NOFDM) case are described. The

whole paper is summarized in the last section.

2. PHY Abstraction Methodology

Physical-layer abstraction methodology for predicting in-

stantaneous link perfo rmance for OFDM systems has been

an active area of research and has received considerable

attention in the literature [ 2]–[11]. T he content in this sec-

tion is based on the evaluation methodology document [1]

of the ongoing work in IEEE 802.16m Task.

rate (BLER) across the OFDM subcarriers transmitting this

forward erro r correction (FEC) coded block for any given

116

PHY Abstraction Methods for OFDM and NOFDM Systems

channel realization (not averaged over the channel statis-

tics). To do that, the vector of post-processing SINR values

at the input to the FEC d e coder are also considered as the

input to the PHY abstraction methodo logy. As the lin k-level

BLER curves are always generated ba sed on a frequency-

flat channel for various SINR’s, an effective SINR (ESINR)

Several ESM approaches for predicting the instantaneous

link performance have been proposed in the literature, in-

cluding: mea n instantaneous capacity [2]–[4], exponential-

effective SINR mapping (EESM) [5]–[8] and mutual infor-

mation effective SINR mapping (MIESM) [9], [10]. Each

of these approaches uses a different function to m a p the

vector of SINR values to a single nu mber. In general, any

ESM PHY abstraction method can be described via the

followin g equation:

)

where: SINR

is th e effective SINR, SINR

in the nth subcarrier, N is the nu mber of symbols in a c oded

block, or the number of subcarriers used in an OFDM sys-

processing per-ton e SINR is therefore dependent on the

transmitter/receiver multiple input multiple output-space

time coding (MI MO-STC) structu re used to modulate /

demodulate the symbols.

2.1. PHY Abstraction Metho ds for OFDM Systems

2.1.1. Mutual Information Based Effective SINR Map-

The computation of the mutual information per coded

bit can be derived from the received symbol-level mu-

SINR

n

where:

β

pends on the mo dulation and coding scheme (MCS) and

the encoding block len gth.

of sub channels). For the selected schemes, 100 different

channel realizations with normalized total power (over the

used subchannels) were produced.

The simulation par a meters are:

Adrian Kliks, Andreas Zalonis, Ioannis Dagres, Andreas Polydoros, and Hanna Bogucka

nels (FUSC);

line are the different realizations of the frequency selective

case. T he figures display the BLER based on the chosen

ESNR metric per different channel realization.

Fig. 1. The B LER based on the RBIR metric, 4-QAM, rate

= 0.75, per 1 subchannel.

Fig. 2. The BLER based on the MMIB metric, 64-QAM, rate =

0.5, per 3 subchannels.

From the system level sim ulation perspective, the perfor-

mance cu rves for all the methods exhibits low dispersion

and good predictio n since they are close to the AWGN ref-

erence c urves. In the EESM case a calibration factor is

used (

β

) for each MCS. The performance is plotted for two

different selections of

. This demonstrates that the pre-

diction accuracy in different BLER areas can be controlled

via proper calibration, and this property can be exploited

in the design of adap tive m odulation and coding (AMC)

algorithms.

3. PHY Abstraction Proposal

for NOFDM Systems – Theoretical

In this section the main fe atures o f NOFDM systems are

118