Research On Ant Colony Algorithm Based On Wireless Sensor Networks

As a technology to gather information, remarkable effects have been achieved for Wireless Sensor Network (WSN), which has broad application prospect in the healthcare, industrial control, and traffic control and other fields due to the characteristics of layout flexibility and self-organization. With the enhancement of people's degree favor to machine Intelligence, a number of routing algorithms based on WSN have emerged. Ant colony algorithm, as a bionic evolutionary approach, by simulating the process of ant colony searching food along the shortest path, which also has robust, versatile, good reliable and other advantages, has been successfully used to solve a series of combinatorial optimization problems that traditional optimization algorithms can hardly be effective, causing widespread concerns in academia and industry. But there exists some problems including long time to search for the optimal path and easy to fall into local optimal solution in the related algorithms, which is researched deeply in this paper.As the research object, ant colony algorithm is mainly discussed. The basic principle of ant colony algorithm is studied at first. According to the current research status, the most serious issue is that no strict theoretical about algorithm parameter values, the effective method of "Trilogy" to determine optimal combination of parameters is proposed in this thesis by studying the number of ants m, inspired pheromone factor a, expect inspiration factor ?, pheromone evaporation factor p.Secondly, node energy consumption is analyzed when ants choose the neighbors to prevent a single node energy depletion phenomenon and balance network energy considering the routing stagnation and easy to fall into local optimal solution. Also, deferred update mode and the negative update on the worst solution is used to quickly narrow the spatial extent and local optimal search strategy 2-opt is applied to swap adjacent edge of feasible solutions, which is increased the chances of local mutation solution to improve the efficiency of the algorithm.Thirdly, a new packet format is designed for network extra energy consumption issues caused by redundancy field of EEABR. The pheromone update way is described to improve the convergence speed for forward ant. Link length and neighbour energy are considered for backward ant so that more pheromones at the place near destination node can be released which also contributes to enhance the likelihood of finding the destination node and promote convergence. The improved algorithm is calculated inside the node which reduces the burden on energy data transmission network. Furthermore, data packet structure is simplified and network energy consumption is decreased combined with the differences task of forward and backward ants.