Loading...
| Friend's email | |
| Your name | |
| Your email | |
| enter code | |
This page was sent successfuly
- Type of Document: Ph.D. Dissertation
- Language: Farsi
- Document No: 51811 (05)
- University: Sharif University of Technology
- Department: Electrical Engineering
- Advisor(s): Bastani, Mohammad Hassan; Amini, Arash
- Abstract:
- Due to multiple transmit-receive channels, the signal model in a MIMO radar system is usually described by high dimensional data structures. However, the desired target space (e.g. range-azimuth domain) which shall be estimated, is mainly sparse (the number of existing targets is usually small). This observation has promoted the use of sparse recovery methods in multi-target detection and estimation in such radar systems which led to introducing the concept of compressive sensing (CS) based MIMO radars. Successful implementation of CS techniques for recovery of radar scenes (for target detection/estimation) from the received noisy measurements strongly entails that the associated sensing matrix could satisfy some specific conditions. In this regard, the restricted isometry property (RIP) provides a guarantee for the recovery performance of a given sensing matrix. Although RIP is a strong sufficient condition for the recovery, its verification for a given matrix is an NP-hard problem. Besides, in radar scenarios, the general form of the sensing matrix is mainly dictated by the physics of the problem and could be marginally controlled using design parameters such as the transmitting waveform. In such cases, the mutual coherence is a common alternative to RIP for designing or optimizing the sensing matrix. In this thesis, considering a CS-based colocated MIMO radar, we attempt to design and modify the sensing matrix through minimizing the coherence. To do so, we employ different tools and degrees of freedom in a MIMO radar including power allocation, waveform design and antenna placement.For power allocation, we derive and solve a convex optimization problem constrained on the total and per antenna maximum power budgets aiming at minimizing the coherence of the resulting sensing matrix. For waveform design, unlike the existing methods which directly attempt to optimize the waveforms and so do not have any control on the practical aspects of the designs, we follow a two-step approach. We first show that the coherence measure depends only on the covariance matrix of the waveforms (rather than the waveforms themselves) and introduce three different convex programs to achieve the covariance matrix. Then, we transform the covariance matrix into realistic waveforms; Due to existence of multiple solutions, we apply additional constraints such as constant modulus (CM) to the signal synthesis process. Simulation results confirm that the introduced designs improve the detection performance of a CS-MIMO radar. For antenna placement, we consider a CS-based colocated MIMO radar with linear arrays. In particular, we choose antenna positions within a given grid. Due to combinatorial nature of the selection problem, we relax it and attempt to find continuous weight values for each location and interpret them as the probability of including an antenna in the given location. Next, we select antenna locations randomly according to the obtained probability distribution. We formulate the problem for the general case of a MIMO radar with independent transmit and receive arrays where we propose an iterative algorithm. For the special case of a transceiver array, the solution is obtained through a convex optimization approach. Our experiments confirms the superiority of the proposed method over the existing methods. We also propose a similar approach for sensor selection in a MIMO radar with planar array for passive target detection. Furthermore, we study the issue of joint optimization of transmit antenna placement and power allocation. Our goal is to minimize the required number of transmitting antennas while guarantying the detection performance via imposing a constraint on the coherence of resulting sensing matrix. We derive this problem as minimizing the l0-norm of a vector subjected to coherence and total power constraints. To solve this problem, we propose two different methods; the first one is based on minimizing the smoothed l0¬-norm (inspiring by the well-known SL0 algorithm) while the second is based on minimizing the l1-norm. Our simulations demonstrate the superiority of the proposed methods for joint optimization and especially the second method over a decoupled scheme of antenna placement and power allocation.
- Keywords:
- Multi-Input Multi-Output (MIMO)Radar ; Sparse Representation ; Compressive Sensing ; Deterministic Sensing Matrix ; Mutual Coherence of Measurement Matrix ; Colocated Multiple Input-Multiple Output (MIMO) Radar ; Power Allocation ; Waveform Design ; Antenna Placement ; Joint Optimization
Digital Object List-
محتواي کتاب - view
Bookmark