Prateek Mohan Dayal prateek[at]iitg.ac.in
_____________________________________________________________________________________
MPOE for MC-CDMA
This work on MC-CDMA has been guided by Prof. U. B. Desai, Professor, Department of Electrical Engineering, IIT Bombay, India and part of it has been carried out at SPANN Lab, IIT Bombay.
Adaptive filtering in Digital Communication has traditionally been based upon the Minimum Mean Squre Error (MMSE) criterion. As it is the Bit Error Rate (BER) and not the Mean Square Error (MSE) that is the deciding factor in a digital communication system, it is imperative that we base our approach upon something that tries to reduce the Average Probability of Error and not MSE. Minimum Probability of Error approach has this desirable characteristic and is thus expected to have better BER performance. Several detectors based on MPOE criterion have already been proposed for OFDM and CDMA and have shown the superiority of the approach.
MC-CDMA is a fairly recent technique that combines the advantages of CDMA and OFDM. Thus we get better frequency diversity than OFDM and at the same time the effect of multipath fading is minimized. MC-CDMA is a strong candidate for high data rate indoor communication and much research has focused on this aspect of MC-CDMA.
Our MPOE detectors for MC-CDMA not only achieve better BER performance than the MMSE detectors, they also exhibit better Near Far Resistance in case of uplink. Near Far Problem is typical of uplink communication where a centralized power control scheme has not been implemented. We have tested these receivers for downlink and uplink cases. In uplink we have also considered the effect of relative user asynchronism on the performance of the detectors. The MPOE detector still outperform the MMSE detector.
One drawback for the MPOE detector is that in weight adaptation mode, the MPOE detectors are exponential complexity although once in the decision directed mode, they are linear complexity like MMSE detector. The solution to this problem is Minimum Conditional Probability of Error (MCPOE) Detectors, which minimize the conditional probability of error instead for the average probability of error and therfore sacrifice performance for complexity. But the performance loss is negligible compared to the reduction in computational complexity. The MCPOE detectors are linear complexity even in the weight adaptation mode. Compare this with MPOE which is exponential and MMSE which is polynomial complexity.
We have published our results in various IEEE conferences. Please follow the link below to take a look at them
About Me Education Research Resume Blog Beyond ECE Favorites Pictures