Loading...

Performance Improvement of MIMO Radars Based on Sparse Representation

Tohidi, Ehsan | 2018

496 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 51307 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Behroozi, Hamid; Nayebi, Mohammad Mahdi
  7. Abstract:
  8. Inspiring by recent developments of Multiple input multiple output (MIMO) communications, MIMO radar has been introduced and MIMO radar advantages such as higher degrees of freedom, improved resolution, and improved estimation accuracy are shown. These advantages have drawn attention of many researchers and engineers toward MIMO radar. On this basis, and during the recent years, in many researches, effect of increasing the number of antennas and pulses on the radar performance have been studied. Although MIMO radar has the aforementioned advantages, the hardware costs (due to multiple transmitters and multiple receivers), high energy consumption (multiple pulses), and computational complexity (due to the complex model and huge number of samples) limit the usage of MIMO radars in large scale networks. On one hand, higher resolution, detection probability, and estimation accuracy is required, but on the other hand, a lower number of antennas/pulses is desirable. Another challenge in MIMO radars is the optimum usage of observations, and coherent processing as a representative of the optimum processing. Due to the complex structure and model of these radars, especially in widely separated case with antennas far from each other, this problem has been rarely considered which causes observations be less effective. Other crucial issues are data rate, required storage, processing load, and the communication bandwidth demanded for transferring data. In fact, the three topics of coherent processing, high cost, and computational complexity and huge data are the fundamental challenges in MIMO radars. In this thesis, these challenges are studied and elaborated. Keeping in mind the low number of targets in the region of interest, sparse sensing based approaches have shown promising performance. In other words, sparse representation of targets provides the opportunity to reduce the MIMO radars problems. Using sparse representation of targets and employing sparse processing, the three aforementioned challenges in MIMO radars are considered as three approaches to look at the problem that we present the model and attend to provide appropriate solutions. In the first approach, a method of coherent processing in widely separated MIMO radar is proposed in which besides improving the efficiency of the observations usage that results in a better target detection and estimation performance, the measurement matrix dimension is reduced for a few order of magnitudes, and also provides the possibility of radar processing applications such as moving target indicator (MTI). Through the second approach, radar costs reduction with the two criteria of estimation accuracy and synthesized pattern is accomplished. In order to reduce radar costs based on estimation accuracy, the Cram\'er-Rao lower bound (CRLB) for two targets is introduced and its advantages are illustrated. Then, using different measures and proving their convexity and submodularity, efficient and effective algorithms for antenna and pulse selection are presented. The other form of this approach is to reduce the number of antennas with the metrics of beamwidth and sidelobe level. In this case, by reforming the problem, reformulation, and employing dynamic programming, an effective algorithm of reducing the number of antennas while keeping the beamwidth unchanged is proposed that achieves the lowest sidelobe level among all the available methods. The third and last approach of this thesis is to present a method of processing in the compressed signal domain which perform the radar task including detection and estimation in the compressed signal domain. We show that the proposed method implies data rate reduction and reduces the required storage and processing load. Finally, the trade-off between cost and performance for all these methods are presented and we show that in some cases, with the price of negligible reduction in performance, radar costs can be reduced hugely
  9. Keywords:
  10. Coherent Processing ; Sparse Representation ; Dynamic Programming ; Convex Optimization ; Multi-Input Multi-Output (MIMO)Radar ; Submodular Optimization

 Digital Object List

 Bookmark

No TOC