Channel Estimation and Equalization for MIMO-OFDM


Orthogonal frequency division multiplexing (OFDM) is an effective technique for high bit rate transmission. It has found widespread applications and is already in many standards (including those of DAB, DVB, DSL, and WLAN). It has recently been proposed for broadband wireless metropolitan area networks (like WiMAX). OFDM is the key technology of the long term evolution of 3G and one of the core technologies of 4G. OFDM avoids intersymbol interference by inserting a guard band (cyclic prefix) into the transmitted symbol. This effectively divides the channel into many narrowband channels over which parallel streams of data are transmitted. Frequency selectivity can now be mitigated using one tap equalizers. For proper operation of an OFDM receiver, it needs an accurate estimate of the channel state information. For rapidly time-variant channels, the receiver faces the additional challenge of performing channel (and data) recovery for each OFDM symbol.

Journal Papers: [4] [5] [6]
Conference Papers: [12] [13] [15] [16] [17] [18] [19] [20] [21] [26] [28]
Patent: [1]
Standard Contributions: [1] - [9]
Projects: [4] [8]

Data-Aided Channel Estimation:

I am interested in developing adaptive/iterative algorithm for MIMO channel estimation and data detection. In my research in this area, I developed receivers for MIMO-OFDM transmission over rapidly block fading frequency-selective channels while minimizing the training overhead. This is done by making a collective use of the structure underlying the communication problem, a structure induced by the channel and data constraints. Channel constraints include time, frequency, and spatial correlation, finite delay spread, and channel sparsity.  Data constraints include the finite alphabet constraint, cyclic prefix or guard-band, error correcting codes, and space-time codes.

A recent problem that I am considering is inter-carrier interference (ICI) estimation and cancellation for OFDM transmission over time-variant channels.

Blind Channel Estimation:

A recent problem of interest is blind channel estimation for OFDM transmission. Specifically, the presence of the cyclic prefix gives an extra observation that can be used for blind data detection. In contrast to many other blind channel estimation techniques, there is no need for averaging over many symbols, and so the technique lends itself to data-detection for OFDM transmission over block fading channels. 

Back to Research Interests