Research Of Real-time Tasks Load Assignment Simulation In Phased Array Radar System

With the development of science and technology, phased array radar becomes more and more multifunctional. The data to be processed increases sharply, thus increasing the difficulty of radar load calculation and aggravating the radar loading task, but not reducing the real-time requirements of the radar system. The results of radar task processed by the original single processor have been difficult to meet the high load task demands on real-time radar.For fast and efficient processing of phased array radar system load task, distributed parallel processing systems are used. With the use of parallel scheduling algorithm assigning the radar tasks to the client paralleling, the task completion time is reduced and real-time requirement is meted; through scheduling algorithm reasonable to schedule tasks, the system resources are full used and the rate of load balancing is improved. Accordingly, phased array radar tasks and scheduling environment are modeled mathematically to optimize the assignment algorithm and particle swarm algorithm. Applying both algorithms to the tasks scheduling respectively, real-time scheduling and load distribution technology are researched. Specific research work is as follows:The mapping relationship between Phased array radar load task and distributed scheduling environment modeling is established. Given the differences between serial and parallel processing tasks, radar tasks are divided into sub-tasks with smaller communication cost which are suitable for scheduling in distributed systems. Data transmission and communications are reflected by the sub-tasks dependencies between them. Build a mathematical model for radar load tasks from the task dependencies, complexity and quantity. Choose the client-server architecture as the scheduling environment and model the clients and the communication network of scheduling environment mathematically. The objective functions and the constraint conditions are established so that the phased array radar load task schedule in distributed heterogeneous system. Mapping relationship between tasks and environment is established as well.An optimized assignment algorithm based on assignment problem and the Hungarian algorithm is proposed. Given the assignment strategies of assignment algorithm leading to idle waiting time of clients being too long or communication cost being too large, meanwhile, the Hungarian algorithm maybe get into infinite loop, thus the optimization assignment algorithm is put forward: in order to balance the contradiction between client idle time and communication cost, taking tasks quantity as the standard instead of number of task; Taking the comparison between the quantity and the threshold of task accumulated as a standard whether assign tasks or not to adjust the load balance rate; Taking task completion time and load balance rate as the goal to avoid the Hungarian algorithm into an infinite loop. It is proved that infinite loop of the Hungarian algorithm is avoided, the task completion time is shortened and the system load balancing is greatly improved by the simulation and algorithms comparison.An improved particle swarm optimization algorithm based on DPSO is proposed. The PSO has a fast convergence and is easy to fall into local optimal solution, so the improved particle swarm optimization is present: initialize the particle according to the probability matrix which is converted by efficiency to increase the diversity of particles; adaptive inertia coefficient by iterations; take the task completion time and load balance rate as a weighted fitness function, then take the fitness value by negative feedback as learning factor, dual adjustment of global search and local search to reduce the risk of falling into local optimal solution. Numerical simulation of the algorithm is proved to reduce the task completion time and improve the system load balance rate. It is effective for the convergence and avoiding falling into local optimal solution algorithm. Meanwhile, the task completion time is shortened and the system load balancing is improved by the simulation and algorithms comparison.System implementation. Simulate the optimize assignment algorithm and the improved particle swarm optimization in a whole radar simulation system, respectively. The effects of optimization algorithm in simulation system are proved.