|
Résumé :
|
In the last decade, the number of mobile users has witnessed an unmatched growth leading to overloading the cellular network core. To keep up with this development of cellular networks, many technologies are being investigated to get to the next generation of cellular networks “the fifth generation” (5G). One of those technologies promises to offload the network core, to improve spectral and energy efficiency, to reduce delay and to maximize overall throughput. All those promises have been made by the Device-to-Device (D2D) communication technology. D2D communication is a direct communication among devices without involving the Base Station (BS). Many challenges face D2D communication technology to fulfill its promises. Allocating resources and controlling power efficiently lead to minimize interferences and maximize overall throughput. In this thesis, we investigate the joint channel allocation and power control problem for D2D communication underlay 5G cellular networks. In this dissertation, we propose the use of a bio-inspired method because of its efficient properties like self-organization, autonomy, scalability and adaptation. The Bee Life Algorithm(BLA) has achieved best results for many problems like job scheduling in cloud computing and packet routing in vehicular ad hoc networks. For those reasons, we have adopted BLA to solve the joint spectrum and power control problem for D2D communication underlay 5G cellular networks. Afterward, we have proposed a new bio-inspired approach called an enhanced Bee Life Algorithm for spectrum allocation and power control in D2D communications (E-BLAD2D). The E-BLAD2D is considered as a population-based metaheuristic, suggesting an initial population generation in the basis of a simulated annealing algorithm. This non-random initialization increases the capability of obtaining promising solutions to reach optimal D2D communications in 5G cellular networks with best spectrum allocation and power control. EBLAD2D is also based on reproduction and food foraging behaviors inspired by bees colony contributing to this optimization process. After that, we proposed another solution for the joint problem based on Matching Algorithm and BLA named Matching Bees Algorithm (MBA). This last proposition consists of using Matching algorithm (one-to-many with externalities) to generate the initial population by optimally allocating resources followed by the use of BLA to reach the best solution. The three proposed algorithms achieve better networks throughput compared to Genetic algorithm (GA) and Particle Swarm Optimization (PSO) that are widely used to solve the joint channel allocation and power control.
|
