Filter Bank Techniques for the Physical Layer in Wireless Communications

Abstract

Filter bank based multicarrier is an evolution with many advantages over the widespread OFDM multicarrier scheme. The author of the thesis stands behind this statement and proposes various solutions for practical physical layer problems based on filter bank processing of wireless communications signals. Filter banks are an evolved form of subband processing, harnessing the key advantages of original efficient subband processing based on the fast Fourier transforms and addressing some of its shortcomings, at the price of a somewhat increased implementation complexity. The main asset of the filter banks is the possibility to design very frequency selective subband filters to compartmentalize the overall spectrum into well isolated subbands, while still making very efficient use of the assigned bandwidth.

This thesis first exploits this main feature of the filter banks in the subband system configuration, in which the analysis filter bank divides the original signal into its subbands and the subsequent synthesis filters reconstruct the original signal in a properly processed form. Here the application is narrowband interference cancellation in a direct sequence spread spectrum system. Using complex modulated filter banks, complex valued data is efficiently processed and the interference notched out without its effect leaking to adjacent subbands. Means for detecting and estimating the location and power of the interferer are presented here, and a novel recursive excision algorithm leads to enhanced interference mitigation performance.

In multicarrier communications, the filter banks are switched into the transmultiplexer configuration, with a synthesis filter bank in the transmitter and an analysis filter bank in the receiver. The high selectivity of the subchannel filters allows them to separate data streams as clusters of contiguous subbands without the need of additional filtering. This raises the question: How much processing can be moved to the low rate subband streams, after the analysis filter bank? Subcarrier-wise processing enables independent processing of data streams allocated to different frequency blocks and separated by the filter bank selectivity. These blocks could correspond to different users in an uplink, with different timing and frequency offsets that are difficult to resolve from the compound high rate signal. Our research objective was to perform subcarrier-wise channel estimation and synchronization, including equalization, as well as narrowband interference cancellation when needed.

At the start of the studies of our research group, subchannel equalization was not well understood. Its analysis resulted in efficient equalization methods and later in techniques to estimate the channel impairments to perform the equalization. In this sense, this thesis presents alternative approaches for estimating channel parameters, such as timing, frequency offset and channel frequency response. One estimation technique is pilot based, suitable for systems in which the channel frequency response is almost flat within the subchannels. The other relies on training sequences, an alternative that fits well in implementations with moderate number of subchannels and mildly frequency selective fading channel within the subcarriers. Both approaches are known for the forerunner in multicarrier communications, OFDM, but here the particular difficulties of the filter bank multicarrier signal are tackled.

The presented techniques result in a well performing and quite complete multicarrier communications solution. Its application in a WiMAX-like setup proved to provide throughput gains of up to 20% compared to the original, OFDM based reference system. Based on the extended research and on the results published in the publications that are compiled as part of this thesis, the author strongly believes that the filter bank approach is the right direction to follow in multicarrier communications.