Development of spectrum sensing cooperation and fusion strategies for tactical heterogeneous networks in mobile environments

Abstract

Spectrum sensing in tactical wireless networks that are under attack by jammers is an important consideration to ensure the safety and effectiveness of deployed military personnel. It is necessary for members of these networks to be aware of which channels in the spectrum are compromised and which are safe to use for data transmission. Wireless transmitters that compose these networks can accomplish this by sensing the energy level on different channels in order to determine if there are jammers active on these channels. They can then share this information with their peers in order to collaboratively identify and avoid jammers. Several solutions based on reinforcement learning exist that allow wireless transmitters to devise a transmission Policy based on their observations of the jammers’ activity, but these solutions often falter when the behaviour of a jammer is random, thereby preventing reinforcement learning algorithms from learning and anticipating their behaviour. In this thesis, we first discuss collaborative spectrum sensing and the theory behind cognitive radios, jamming, and anti-jamming. Next, we detail the system model used to represent the multi-agent anti-jamming problem. We then introduce a collaborative pseudo-random channel selection algorithm and a data collaboration and fusion scheme based on super-decision vectors in order to improve awareness of spectrum utilization across the network. Simulation results show that this solution leads to higher rates of detected jammers as well as an increase in the number of transmissions occurring on unjammed channels.

Date	2 Apr 2020
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Georges Kaddoum (Supervisor)