TURBO TRELLISCODED MODULATION FOR DATA TRANSMISSION ON 25 kHz VHFUHF CHANNELS

1
TURBO TRELLIS-CODED MODULATION FOR DATA
TRANSMISSION ON 25 kHz VHF/UHF CHANNELS

• Michal Sybis
• Chair of Wireless Communications
• Poznan University of Technology

2
Outline

• Motivation of the work
• Introduction to Parallel and Serial Turbo-TCM
• Decoding algorithms MAP i SOVA
• Interleaver design (S-random and CMI)
• Simulation assumptions
• Simulation results
• Conclusions

3
Motivation

• The purpose of the project is to research
  radio-modem that fulfill the following condition
• transmit data or digitized speech,
• terminals are able to move at the speed of up to
  100 kmph,
• transmission is ensured in the range from 5 to 10
  km in the open environment,
• carrier frequencies can be selected in the range
  between 25 and 512 MHz,
• Software Defined Radio (SDR) implementation.

4
Turbo-TCM

• Turbo-TCM versus classical turbo-coding
• TCM encoders are applied instead of CC encoders,
• the interleaver operates on a group of m bits
  instead of single bits,
• achieving the desired spectral efficiency
  requires mandatory puncturing of the parity bits
  and is not quite as straightforward as in binary
  turbo-codes,
• there are some restrictions on component codes
  and interleavers.

5
Parallel Turbo-TCM
TTCM encoder parallel concatenation
6
Parallel Turbo-TCM
TTCM decoder parallel concatenation
7
Serial Turbo-TCM
TTCM encoder serial concatenation
8
Serial Turbo-TCM
TTCM decoder serial concatenation
9
MAP algorithm

• The purpose of the algorithm is the calculation
  of a posteriori probabilities
  or .
• Summing over all transitions
• where probabilities of a specific transition in
  the trellis are

10
MAP algorithm
Forward recursion
Backward recursion
Branch metric
11
SOVA algorithm

• SOVA algorithm estimates the soft output
  information for each symbol and is based on the
  Viterbi algorithm. The decoder output values,
  i.e., soft decisions are defined as

Each value is calculated according the
formula
12
SOVA algorithm

• Decoding process
• forward and backward recursion metric
  calculation,
• calculating soft decisions

The path metrics are computed according to an
ordinary formula
The branch metric is calculated according the
formula
13
Interleaver design
The interleaver must map even positions to even
positions and odd ones to odd ones. Transmitted
symbols are chosen alternately from the first and
second TCM encoders. Even-even interleaver.
14
Interleaver design (S-random)

• It can be said that an interleaver has spreading
  factors (S, T) if

To keep the reasonable search time parameters S
and T have to satisfy the condition


In a practical implementation we usually choose S
T.
15
Interleaver design (CMI)

• Equations that have to be satisfied in CMI

where
and K is the maximum number of FP to be
eliminated and can be calculated from the formula
16
Simulation assumptions

• Constellation 16QAM
• Coding efficiency R 0.75 for parallel and R
  0.5 for serial concatenation Turbo-TCM
• 8-state TCM encoders for parallel TTCM
• 8-state TCM inner encoder and 4,8,16-state outer
  encoder for serial TTCM
• AWGN channel and static two-path channel model
  (delay of the 2nd path is 13.3 µs, relative
  attenuation of the 2nd path is 8.6 dB),
• Square-root raised cosine pulse shaping filter
  with the roll-off factor of 0.25 is applied
• Symbol rate 21500 symb/s, (information bit rate
  64500 b/s)
• Perfect synchronization.

17
Simulation assumptions
Simulation assumptions

• Investigated receiver configurations
• Serial TTCM
• Parallel TTCM
• LE parallel TTCM
• DFE parallel TTCM
• LE serial TTCM
• DFE serial TTCM

18
Results
BER results for the serial TTCM scheme, N  512,
AWGN channel
19
Results
BER results for the serial TTCM scheme, N  2048,
AWGN channel
20
Results
BER results for the parallel TTCM scheme in AWGN
and fading channels
21
Results
BER results for the serial TTCM scheme with S  8
in AWGN and fading channels
22
Results
BER performance comparison for various
interleavers with size N 512
23
Conclusions

• For the simulated channels performance of LE and
  DFE is very similar
• Serial TTCM has a better performance for high SNR
  (lower error floor)
• TTCM with Code Matched Interleaver achieve better
  performance than with S-random interleaver
• Further intensive simulations are necessary
• extension on time varying channels
• turbo-equalization and turbo-synchronization