Research On The Cloud Computing Resource Allocation Based On Improved Ant Colony Optimization

With the extensive application of information technology in various fields, the progressin Internet technology and increase in the number of Internet users have led to an explosivegrowth in data traffic. Internet has now entered the "era of big data". The traditionalcomputing model is no longer able to meet the currently dynamic, changing needs. It is insuch a development context that "cloud computing" came into being, making itself the thirdIT wave following the personal computer revolution and the Internet revolution. Cloudcomputing is the further development and commercial realization of Distributed Computing,Parallel Computing, Grid Computing, Network Storage and Large-scale Data Center.In the development process of cloud computing, the resource allocation issue, as one ofits key technologies, is still in the research stage. In the entire ‘cloud’ environment, theresource and structure distribution as well as its realistic situation is very complex. There canbe many unpredictable significant changes in the network load of any line at any time.Possible underestimation of demand for resources will lead to the fact that the currentresource scale is not large enough to satisfy the user’s needs, while possible overestimation ofdemand for resources will make some of the resources rented idle, hence greatly decreasingthe resource utilization rate. Therefore, with the increasing number of users and theirever-changing requirements as a premise, how to allocate the computing resources rationallyin an efficient and low-cost way becomes a key issue for cloud computing. And since the coreof resources allocation is the allocation algorithm, the paper therefore tried to provide anoverview of the current resource allocation algorithm for cloud computing and eventuallycame up with an improved resource allocation algorithm based on this.By analyzing the disadvantages in traditional ant colony algorithm in solving the cloudcomputing resources-related problems, an improved method was provided, including theimprovement in the algorithm transition probability and that in the updating methods of localand global pheromone. The innovation of this paper lies in that dynamic factors are included in defining the transition probability and updating the pheromones, which allows thealgorithm to adjust itself dynamically with the increasing iterations. This can help ensure abetter search performance.Meanwhile, considering the blindness of ant colony algorithm in the early iterations, asuggestion was proposed that genetic algorithm and ant colony algorithm be combined in use.In light of fast convergence characteristics given by genetic algorithm, it could be used toobtain an optimum solution, which would then be converted to be the initial pheromone of antcolony algorithm so that an optimized ant colony algorithm could be used for search. In themeanwhile, mutation of the genetic algorithm would be also introduced to realize thevariation of pheromone concentration in order to avoid the local optimum. At last, asimulation platform was borrowed to validate the proposed algorithm.