VLSI Digital Filters

Our research in this area has been directed towards design of concurrent algorithms and architectures for VLSI digital filters. In 1960s and 1970s, digital signal processing algorithms were implemented using the available microprocessors which executed the algorithms sequentially. Therefore, there was no motivation for designing concurrent signal processing algorithms, which could exploit pipelining or parallelism. The advances in VLSI technology and parallel processing has changed our thinking style for signal processing algorithm design. Our research efforts in designing concurrent algorithms have been directed towards transforming existing non-concurrent algorithms into concurrent forms as well as designing new algorithms which are inherently concurrent. We have developed look-ahead transformation, decomposition, and incremental computation techniques to create pipelining and parallel processing in signal processing algorithms. In addition, we have proposed sum, delay, and product relaxed look-ahead techniques to design inherently pipelined adaptive digital filters. We have addressed design of pipelined and parallel recursive digital filters, recursive lattice digital filters, recursive wave digital filters, LMS adaptive digital filters, adaptive lattice digital filters, two-dimensional recursive digital filters, and rank-order and stack digital filters. We have also examined finite word-length effects in these filters for fixed-point hardware implementations. Our approaches introduced pipelining in these algorithms. Pipelining of recursive digital filters was considered impossible before. For the first time, we demonstrated feasibility of pipelined stable recursive digital filters. For the first time, we also introduced pipelined architecture topologies for various forms of adaptive filters. For recursive least square (RLS) adaptive filters, we showed that these structures can be easily pipelined by using our proposed Scaled TAngent Rotation (STAR) rather than the common Given's rotation. The STAR rotation is approximately orthogonal as opposed to exactly orthogonal. Recently we have demonstrated that by using Annihilation Reordering Look-Ahead recursive least square (RLS) adaptive filters can also be pipelined based on Givens rotation; these filters maintain exact orthogonality. Truly orthogonal IIR recursive filters have also been developed. These structures provide excellent roundoff noise properties.
 


Selected Publications


[Back page]  [Prof. Parhi's homepage]