New directions for Next Generation Array Signal Processing
Abstract
Array signal processing (ASP) has repeatedly witnessed major new developments every fewyears. In recent years, there has been renewed momentum in this area because of increased activity in5G networks. For example, because of the properties of mmWave bands in 5G, it is very important todevelop new ways to perform highly effective beamforming to connect individual users. Research onlarge arraysis becoming important, both in terms of exploiting their full potential, and decreasing themassive computational complexity at the outputs of such arrays. The focus of this proposal is basedon exciting new ideas for next generation array signal processing research. One of these is inspiredby the recently introduced notion of convolutional beam space (CBS) in array processing, which offerscomputational efficiency crucial for large arrays. One proposed focus is the use of the CBS platform fora new generation of multistage beamformers and DOA estimators especially suited for large arrays. Theextension of convolutional beamspace to the case of sparse arrays will also be considered here. In thearea of distributed or decentralized array processing, which has been of interest for some years, thereare a number of unsolved problems of significant interest, which will be one of the focus areas here.For example, distributed implementation of sparse arrays with large coarrays, distributed implementationof convolutional beamspace, and distributed optimization of many array processing algorithms will begiven attention. Sparse arrays, which have seen intense activities in the last ten years, continue to bean important research area, for example, in regard to the robustness of the difference coarray to sensorfailures. The importance of this arises from the fact that good sparse arrays derive their advantages froma good difference coarray. If the coarray contains a large uniform linear array (ULA) as a subarray, thatwill lead to powerful methods for estimation of angles of arrival, often more angles than the number ofphysical sensors. Thus, preserving the coarray in the presence of inevitable element failures is important.A general platform for the formulation of robust problems for sparse arrays will be developed. Sparsearrays which minimize the number of element pairs with the smallest spacing (half-wavelength spacing)are of significant importance in reducing mutual coupling, and there are several research problems inthat area that will receive attention in the proposed work. Another important problem that will receiveattention is the problem of denoising a periodic signal. While this has seen some history, the focushere will be a new type of analysis/synthesis approach. Namely, an effective combination of Ramanujananalysis filter banks and Ramanujan or Farey synthesis dictionaries will be considered for this problem.This will open up important new ways to denoise periodic signals. In particular, it will offer the uniquefeature that the denoised signal is exactly periodic, which is not a property shared by classical algorithms.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jun 09, 2021
- Source ID
- N000142112521
Entities
People
- Palghat Vaidyanathan
Organizations
- California Institute of Technology
- Office of Naval Research
- United States Navy