- مبلغ: ۸۶,۰۰۰ تومان
- مبلغ: ۹۱,۰۰۰ تومان
The development of a powerful search mechanism to find a good solution is the current research direction of studies on metaheuristic algorithms; however, most of the developed mechanisms will search and check the possible solutions without knowledge of the overall landscape of the solution space during the convergence process. To make each search during the convergence process as effective as possible, this paper presents a new metaheuristic algorithm called search economics (SE) to solve the deployment problem of wireless sensor networks. The main distinguishing features of the SE are twofold: the first is its capability to depict the solution space based on the solutions that have been checked by the search algorithm, and second is its capability to use the knowledge thus obtained, i.e., the “landscape of the solution space,” during the search process. On the basis of these concepts, the investment in a search process will be more meaningful and thus less easy to fall into a local optimum during early iterations. The experimental results show that the proposed algorithm can provide a result for the deployment problem that is significantly better than those provided by the state-of-the-art metaheuristic algorithms evaluated in this study in terms of the quality.
This paper presents an effective method called SE to enhance the performance of metaheuristic algorithms by keeping track of the information collected from the search process, which includes not only the objective values of the candidate solutions but also the parts of the solution space that have been explored. The solution presented in this study is to depict the solution on the basis of the information we have in hand so that the search algorithm can dynamically invest computing resources in a region in the solution space that has a higher potential to find a better solution, meaning that SE will do its best to invest the computing resources it has in the regions where a better solution can be found. With the addition of computing resources (computer nodes) to the search process, the proposed algorithm will divide a region into two new regions by design and then assign the searcher belonging to the original region and a new searcher to these two new regions to achieve the goal of parallel computing on the fly. The simulation results of the OneMax problem and DP show that the proposed algorithm is able to find better results than the other metaheuristics compared in this paper. Some of the simulations further show that the proposed algorithm is able to find the global optimum if we invest enough computing resources, which implies that SE will not easily fall into a local optimum. Although it is unable to store all of the searched solutions for the time being, SE now uses lossy compression to keep track of all of the searched solutions (i.e., not just the best-so-far solution but also the others). However, the ultimate goal of SE is to use lossless compression to keep track of all of the searched solutions. This is why this paper can be regarded as the starting point of this study. In addition to developing a better way to depict the solution space, the goal is to apply SE to other optimization problems in the future to demonstrate the performance of the proposed algorithm.