RCML Algorithm for CDMA Systems

1
Optimum Multiuser Detection in CDMA System
Fatih Alagoz
2
Outline
Code Division Multiple Access (CDMA) System
Model Problem statement and motivation Optimum
multiuser detection. The proposed algorithm for
CDMA System complexity and performance measures
in AWGN Channel. Conclusion future work.
3
Multiple Access Communication Systems

• Frequency Division Multiple Access- (FDMA)
• Time Division Multiple Access- (TDMA)
• Code Division Multiple Access (CDMA)

4
CDMA System Model
Multi-paths
1
Multi-paths
K
1 billion mobile users !!! US 100 billion/year
!!!
5
Problem statement motivation

• Optimum multiuser detection (find optimum using
  exhaustive search algorithm) i.e., 2K
  computational complexity in the of users, K.
• Exceptions with polynomial complexity stringent
  requirements on the signature waveforms design.
• These requirements limit the system capacity.
• Motivation Design optimum/suboptimum detectors
  with acceptable complexity and performance.

6
CDMA in AWGN Channel (1)
The received signal employing antipodal
signaling where K number of users, Ek
Energy/bit for user k, sk(t) unit-energy
signature waveform for user k, bk?1,-1 bit
value for user k, T bit interval, n(t)
Additive White Gaussian Noise (AWGN) with
one-sided power spectral density No.
7
CDMA in AWGN Channel (2)

• The output of K filters matched to the users
  signature waveform and sampled at T are
• where
• The output of the matched filters are sufficient
  statistics for the optimum detector

8
The Idea
Aim is to reduce computational complexity while
maintaining the optimum detection
View the coefficients of the optimum metric as
weights indicating the order in which the bits
are estimated
Achieve decision regions to reduce the complexity
while providing optimum detection
No-need to compute the insignificant terms !!!
9
Reduced Complexity MaximumLikelohood (RCML)
Algorithm (1)

• It is based on the Maximum Likelihood (ML)
  metric
• It views the coefficients of the bits in the ML
  metric Ai, Bij, ?i?1,Kand jgti as weights
  that indicate the order in which bits can be
  estimated.
• Large values of the coefficients have more
  effect on deciding the bit value than smaller
  values, i.e. Order of their contribution to the
  ML metric.

10
RCML Algorithm (2)
bn is optimum solution iff
Example K3,
11
RCML Algorithm (3)
if
Rule.1
Rule.2
if
elseif
PRUNE
end
Rule.3
Once User i is optimally detected,
apply the rules to K-1 user system.
12
A Few Results Complexity
C o m p l e x i t y
Blue Optimum Red SDP Green RCML
Number of Users (K)
13
A Few Results Average Bit Error Rate (BER)
in lightly loaded CDMA Systems
B E R
Signal to Noise Ratio (Eb/No) in (dB)
14
A Few Results Average Bit Error Rate (BER)
in highly loaded CDMA Systems
B E R
Signal to Noise Ratio (Eb/No) in (dB)
15
Expert Comments... for the
Proposed RCML Algorithm
Complexity is lower than that of SDP Algorithm
and significantly lower than ML (Optimum)
Algorithm BER performance is better than SDP
algorithm and similar to ML algorithm
16
Whats Cooking Next ?

• Extend the RCML algorithm for Devising a New
  Suboptimum Multiuser Detector
• Consider coefficients that are greater than some
  certain value Z (eg. mean).
• Terminate the algorithm if the largest value
  does not change after P stages.

Test the performance of algorithms for
Asynchronous CDMA systems
Extend the RCML algorithm for fading channels
17
Please Read
F. Alagoz, A New Algorithm for Optimum
Multiuser Detection in Synchronous CDMA
Systems, Int. J. of Electronics Commun.,
vol. 57, 2003.
F. Alagoz, and A. Al-Rustamani A new branch
and bound algorithm for multiuser detection,
Proc. of Int. GAP Conference, Turkey, June,
2002.
F. Alagoz, and M. Abdel-Hafez RCML Algorithm
for Suboptimum Multiuser Detection in CDMA
Systems, in prep. IEEE Trans. on Commun. (end
of 2003).
18
Acknowledgements.

• Dr. P. Tan of Chalmers University, Sweden, for
  providing the material on SDBP algorithm
• Dr. A. AlRustamani of Dubai Internet City, UAE,
  for her collaboration in Algorithm 1 and 2.
• My colleague Dr. M. Abdel-Hafez of Electrical
  Eng. Dept., UAEU, for his constructive criticism.
• My Dear Students Haifa Abdulla, Muna Alawi, Amna
  Rashid, Sally Asmar and Dina Nasr.
• Finally, The UAE University Research Affairs for
  their trust at the proposal stage of this work...
  and off course, their
  financial support during the course of the
  research....

19
Feel Free to Contact Me
ltltltltltlt Any Questions gtgtgtgtgtgt
20
Extra1 Simplified form of the metric for
Asynchronous CDMA System
21
Extra 2 Example of RCML detection for K3 users