Two Types Of Network Optimization Problems

Network optimization is widely used in many fields such as communication network, computer network, economy management and transportation. Many specific models are defined. The optimization of network coding links and network flow monitoring are the specific models in the field of communication and computer networks. Thus, it is of great value to research in both theory and economy. We do many researches about these two problems in this paper.Contraposing the problem of optimizing network coding links, a mixed optimization algorithm is put forward in this paper, which combines the differential evolution algorithm and the genetic algorithm in hybrid coding. In the whole evolution process, the parental individuals first use mutation and crossover operation of standard differential evolution algorithm to create intermediate group. This can avoid sensitive to the quality of initial population like genetic algorithm which leads to easy to fall into local convergence. Then use roulette selection operation of genetic algorithm to create new population. Thus can largely avoid the possibility of eliminating potential individuals which may occur when use greedy selection operation of differential algorithm. The simulation experiment also proves that this mixed optimization algorithm can be applied to optimizing network coding links and get remarkable results. The innovation point of this algorithm lies in applying the differential evolution algorithm which is used to solve continuous problem to such a discrete problem and combining it with the genetic algorithm to make up the defects of theirs.Network traffic monitoring is an important part of traffic engineering. However, the collection of massive data from large numbers of devices adds more pressure on the network, resulting in the decrements of routers’ performance and increase of costs. To reduce the influence of monitoring and the extra expenses, the deployment of monitoring points is of great importance. In our paper, the problem of deploy monitoring points is abstracted to minimum vertex cover problem. Two vertex selection strategies are adopted in this paper. Considering that solving the minimum vertex cover problem is NP-hard, we proposed a method to eliminate the redundancy, further optimizing the existing results. Now we can get satisfactory solution set by each of the two approximate algorithms. It’s proved effective to many graphs by some specific examples and can obtain better results than other existing algorithms proposed.