Applications of fireworks-based evolutionary algorithms for computationally challenging network problems

This thesis covers two types of contributions: formulation of network optimization problems and algorithms to solve these optimization problems. We propose resource assignment problem in Internet of Things network (IoTN) with three nodes: IoT, core cluster node (CCN) and base station (BS). The assignment of resources, such as CPU and memory, from IoTs to CCNs, and CCNs to BSs is a challenging task. The objective of the problem is to minimize the weighted sum of computational power at CCNs and transmission power between IoTs-CCNs and CCNs-BSs radio connections. We also propose a broadband wireless network (BWN) wherein the planning of BSs, relay stations (RSs), and their connections to subscribers minimizes the overall (i.e., weighted sum of the hardware and operational) cost of the network and reformulate a virtual machine (VM) placement to minimize power consumption in a datacenter. The (re)formulated problems are integer programming problem and finding optimal solutions for these problems by using exhaustive search is not practical due to demand of high computing resources. The practical approach is to minimize power in IoT network and VM placement, and plan broadband wireless network using population-based heuristic algorithms. We propose swarm intelligence-based algorithms, that is, two versions of the discrete fireworks algorithm (DFWA) and its variants. The performance of these new algorithms is compared against the low-complexity Biogeography-based Optimization (LC-BBO) algorithm, the Discrete Artificial Bee Colony (DABC) algorithm, and the Genetic Algorithm (GA). Our simulation results and statistical test demonstrate that the proposed algorithm can comparatively find good-quality solutions with moderate computing resources.