دانلود رایگان مقاله الگوریتم ایمنی بی نظم قابل کنترل برای مسیریابی در شبکه DiffServ

abstract

An integrated routing risk model is constructed, which takes into account the effects of unicast routing on DiffServ network risk consisting of the impacts of interrupted services on network users and path availability. With the objective of minimizing integrated routing risk, a novel controllable chaotic immune routing algorithm (CCIRA) is proposed. Due to the inefficiency of traditional path generation methods, a path generation method based on chaotic search and dynamic adjacency matrix is proposed, improving the generation effi- ciency of available solutions of routing optimization algorithms. An evolutionary strategy which combines dynamic vaccination and free mutation is used in order to ensure the population diversity and the global convergence of CCIRA. Chaotic search is introduced to population initialization, vaccination and free mutation in order to overcome the uncertainty of the optimization process and optimization results in traditional evolutionary algorithms due to the crossover and mutation strategies being based on random numbers. Simulation results prove that CCIRA is highly efficient and practical. Combining the integrated routing risk model and CCIRA, the risk control performance of our risk-aware routing algorithm is also proved to be superior by the comparison with other algorithms.

6. Conclusions

In this paper, taking into consideration the effect of service routing on network service layer, transport layer, and physical topology layer risk, we created an integrated routing risk model, after which we proposed a controllable chaotic immune routing algorithm (CCIRA) in order to reduce the routing risk. Due to the inefficiency of traditional path generation methods, we proposed a path generation method based on chaotic search and dynamic adjacency matrix. This method can efficiently generate feasible solutions, improving the efficiency of routing optimization algorithms. In CCIRA, the use of a method that combines dynamic vaccination and free mutation ensures the convergence rate and global optimization capability. The use of chaotic search strategy instead of probability-based strategy during the path generation, vaccination, and free mutation stages improves the controllability and practicability of the algorithm because of the pseudorandomness, ergodicity and determinacy of chaotic sequences. On the LATAX and ITNA networks, the outstanding optimization performance of CCIRA is proved by the simulation results. The risk control performance of the combination of the integrated routing risk model and CCIRA is also proved to be superior by the comparison results with the other risk-aware routing algorithm. This paper conducted an exploratory study on the feasibility of using chaotic search strategy instead of probability-based strategy in evolutionary algorithms for routing problems, and can be used as a reference for future research on improving the controllability of evolutionary algorithms for routing problems.