Grid Task Scheduling Algorithm Based On Adaptive Fault-tolerant Mechanism

The grid computing as a heterogeneous and dynamic platform, its task scheduling goal is to obtain QOS, precedence constraints, performance index and make sure that the task will be executed in parallel with reasonable allocation strategies in the corresponding resource nodes reasonable and orderly. Different from traditional grid scheduling, task scheduling in wireless network grid is more complex, because of WSN has limited battery power, narrow communication bandwidth, data rate, and how to deal with the problem of different applications' failure with adaptive fault-tolerant is still a challenge. So how to combine the existing task scheduling algorithm with wireless sensor grid to achieve optimal time span according to certain strategies and obtain adaptive fault-tolerant ability as far as possible is becoming a very meaningful research direction.This paper introduces the wireless sensor network and the grid is how to integrated, introduces the significance of wireless sensor grid, and put forward the task scheduling in wireless sensor grids divided into two stages:data collection phase and calculation stage. Sensor network needed to complete the task of collection all the raw data, using the grid technology to calculate the raw data and finally complete the task, data collection phase need to be finished before analysis and calculation stages. So can effectively reduce the WSN nodes'energy consumption, make sensor node used only to collect and transmit information, and effectively using the grid technology processing all computing tasks, increase calculation efficiency and accuracy. In grid computing, this paper puts forward REP_GA_TS Scheduling algorithm, critical task will be adopted in multiple resources nodes, according to its priority, so can effectively reduce the transmission delay between the correlation tasks. GA_TS can effectively overcome the GA scheduling algorithm premature convergence,the poor quality and can effectively improve the efficiency of TS scheduling algorithm. In WSN, this paper puts forward the graphical neural mode matching algorithm, GN mode matching can effectively reduce the sensor node context-aware, data transfer and other tasks energy consumption and prolong node life. When error accord we used the task scheduling algorithm of REP_GA_TS, according to the error level to maximum ensure the successful mission execution. This paper analyzes the common grid simulator and mainly introduces GridSim simulation tools, finally through a series of simulation experiment to verifies the GA_TS, REP_GA_TS task scheduling algorithm and adaptive fault-tolerant strategy was effective and scheduling efficiency.This paper still has some problems need to do in-depth research:how to construct a perfect task scheduling strategy for grid and wireless sensor network is the focus of research in the future. In addition REP_GA_TS task scheduling algorithm is difficult to determine Genetic algorithm convergence conditions, and the research emphasis in the future will be to design a reasonable threshold, that the genetic algorithm in the right time to stop, to improve the efficiency of the algorithm.