I have recently focused my research on two specific areas in multiple-input, multiple-output (MIMO) wireless communications: (a) linear precoding in multiuser MIMO downlink communications and (b) the role of cooperation in large scale wireless networks. MIMO systems avail of multiple transmitting and/or receiving antennas. In general, my focus is on developing practical signal processing algorithms and other schemes, e.g., low computation load algorithms, focusing on linear processing. This page also describes some past projects undertaken.

Please visit the publications page for more detailed information.

Linear Precoding

Precoding deals with the area of transforming signals to be transmitted to best match the channel between the transmitter and receiver. Precoding, therefore, requires some information regarding the channel, either its state itself or its statistics. Precoding at the transmitter may be combined with adaptive processing at the receiver. Linear precoding deals with linear transformations specifically. Some specific issues we are addressing are:

■   developing flexible linear precoding for transmitters with an arbitrary number of antennas, communicating simultaneously with multiple users each with arbitrary numbers of antennas, each (within limits) receive an arbitrary number of data streams

■   extending the work above to several figures of merit, such as mean squared error, maximizing data rate etc.

■   developing effective schemes depending on what information is available: channel statistics, channel state information, channel state information with error due to limited feedback or delay, etc.

■   extending all of the above to MIMO-OFDM systems

Cooperation in Wireless Communications

Since the benefits of MIMO communications are now well-accepted, recent research has attempted to extend these benefits to wireless nodes with only a single antenna. Cooperation allows for such nodes to share their antennas to form a MIMO systems. Cooperation can be analyzed from several angles. We are focused on two types of networks: (a) sensor networks comprising low-cost, battery-operated nodes using cooperation to maximize battery life and (b) networks of access-points that service individual users using cooperation to maximize data throughput and reliability.

Both kinds of networks comprise many, many, wireless nodes, though research in the physical layer has generally focused on networks of three nodes: a data source, a data sink (the receiver) and a cooperating relay node. Some of the issues we are addressing are:

■   developing a understanding of the fundamental limitations of cooperation in large scale networks

■   extending the analysis of the 3-node cooperation schemes to large scale networks

■   developing cross-layer approaches that enable cooperation in large-scale networks

■   extending concepts from error control coding to cooperative distributed networks

This work has been supported by Bell University Laboratories and NSERC (via a CRD grant and a Discovery Grant) and in part by a BUL grant.

Past Projects

■   Blind channel estimation in MISO systems

This projected developed an algorithm, based on fourth order cumulants, for blind channel estimation in systems using orthogonal space-time block coding. The innovation here is that blind channel estimation is developed for systems using multiple transmit, but only a single receive, antenna. A particularly important application is the popular Alamouti scheme.

(Sponsored by Bell Mobility and Samsung of Korea)

■   Joint spatial-temporal adaptive signal processing for wireless communications

This project developed an approach with low computation load combining spatial and temporal processing in uplink CDMA systems. The algorithm is based on the joint domain localized (JDL) processing algorithm designed for radar systems. An especially interesting aspect of this algorithm was the extension to multi-cell CDMA systems. This extension is based on the hybrid JDL algorithm wherein an adaptive zero-forcing stage is followed by another adaptive minimum mean squared error processing.

 (Sponsored by NSERC and Bell University Laboratories)

■   Position location in wireless communication networks

The practical development of position locaton techniques, in indoor, cellular and ad-hoc networks will enable many value added services, such as E-911, local map and entetrainment information, etc. Continuing in the theme of practical signal processing, this research effort looks developing practical position location techniques, especially for CDMA based wireless communications. One project under this effort has investigated the impact of and compensation for mutual coupling between the elements of an antenna array in a CDMA setting. The results confirm that mutual coupling is an important phenomena and develops new schemes to compensate for this coupling.

We also developed the Matrix Pencil algorithm for low computation load position location based on both direction of arrival (DOA) estimation and time of arrival (TOA) estimation. In TOA estimation, the Matrix Pencil approach is especially useful since it requires only a single data measurement before estimation can be performed.

(Sponsored by Nortel Institute for Telecommunications)