Non-myopic Scheduling Under Multi-Sensor Collaborative Tracking

The core of sensor scheduling is to study how to control sensor parameters.On the basis of information fusion,the overall performance of multi-sensor networking is improved by rationally scheduling the limited resources of multi-sensor networking.The multi-sensor scheduling strategy is divided into short-term(Myopic)and long-term(Non-myopic).The Non-myopic sensor scheduling pays more attention to the long-term performance of the system.The metric is based on the cumulative revenue generated by scheduling the action sequence in a time domain.This thesis focuses on the Non-myopic sensor scheduling problem of sensors under multi-sensor cooperative tracking,and uses the idea of rolling time domain optimization in predictive control to optimize sensor operating parameters dynamically and in real time.Some effective multi-sensor cooperative tracking Non-myopic sensor scheduling methods are proposed.1.The Non-myopic sensor scheduling of sensors under multi-sensor cooperative tracking with pure angle measurement is studied.Based on the idea of rolling time domain optimization in predictive control and considering the influence of local decision on Non-myopic tracking performance,a Non-myopic scheduling based on single-step decision and multi-step decision-making is proposed.By optimizing the motion trajectory of the movable sensor and finding the optimal multi-sensor coordination scheme,the optimal target tracking performance can be obtained.The simulation results show that in the appropriate prediction step,the target tracking performance under Non-myopic scheduling is better than myopic scheduling.2.The Non-myopic scheduling under the coordinated tracking of airborne dual-base external radiation sources is studied.Considering the long-term impact of single-step decision-making on future tracking performance,a long-distance scheduling model of UAV(airborne receiver)based on multi-step decision-making is established.At the same time,a series of practical constraints such as the motion mode of UVA,collision avoidance and the distance of communication are also considered.A multi-constrained nonlinear optimization algorithm is proposed.Real-time rolling optimization of UAV path for Non-myopic scheduling model improves tracking performance.Finally,a large amount of simulation tests on the proposed algorithm are carried out.3.The distributed MIMO radar sensor resource scheduling is studied.Firstly,aiming at the optimal scheduling problem of distributed MIMO radar source power parameters,a heuristic algorithm for improving convex relaxation is proposed.The algorithm assigns the uniform power as the initial value and optimizes the power allocation through the fast power adjustment strategy.The convex relaxation parameters are calculated and the convex relaxation model is modified.The above process is iteratively performed until the error caused by the relaxation is eliminated,and the relaxation parameters are modified by the fast power adjustment strategy and iteratively to realize the optimal allocation of the distributed MIMO radar under limited power conditions.Secondly,the problem of MIMO radar receiver path optimization is studied.A Non-myopic scheduling model based on multi-step decision-making for receiver collaborative path optimization is established.The localization performance is further improved by controlling the motion parameters of the receiver.