Date of Award

6-2-2023

Publication Type

Thesis

Degree Name

M.A.Sc.

Department

Electrical and Computer Engineering

Supervisor

Majid Ahmadi

Rights

info:eu-repo/semantics/embargoedAccess

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Abstract

Today, due to the increase in the number of users and the need to provide high-rate multimedia services, cognitive radio techniques can be a promising practical solution. In cognitive networks, the unlicensed (secondary) users sense a spectrum utilized by the licensed (primary) users. Due to the interference of the primary users, fast and reliable spectrum sensing is an important challenge in cognitive radio networks. In this dissertation, the resource allocation problem is investigated under min-max optimization framework for establishing fair energy efficiency in wireless sensor networks (WSNs). In fact, we study the sensor selection and power allocation problem in a WSN to minimize maximum energy consumption among nodes. Since the formulated problem is a non-convex and discrete optimization problem, the exhaustive search algorithm can be applied to solve it. Because the exhaustive search is a high complexity algorithm, we propose algorithms with low complexity to solve the problem based on some relaxations and convex optimization methods. In fact, we convert the formulated problem to two sub problems: sensor selection (first sub problem) and power allocation (second sub problem). In the first sub problem, the discrete optimization problem is relaxed to a classical optimization with continuous optimization variables. Then the relaxed problem is solved by convex optimization methods, which derive a cost function for selecting sensors based on priority. On the other hand, solving the second sub problem with the help of convex optimization leads to the transformation of the optimization problem into a one-dimensional search problem. Finally, to find the joint solution, we propose an algorithm that has low computational complexity compared to the exhaustive search. In the following, we present the neural network approach to solve the formulated problem. In this approach, a feedforward neural network is designed to classify sensors into two classes (active mode and idle mode) in the joint problem. The simulation results show that the proposed methods and algorithms outperform the conventional benchmark methods in the energy efficiency literature.

Included in

Engineering Commons

Share

COinS