Exploiting prospect theory and risk-awareness to protect UAV-assisted network operation

• Main Authors: Panagiotis Vamvakas, Eirini Eleni Tsiropoulou, Symeon Papavassiliou
• Format: Article
• Language: English
• Published: SpringerOpen 2019-12-01
• Series: EURASIP Journal on Wireless Communications and Networking
• Subjects: Risk-awareness; Malicious users; Behavioral modeling; NOMA; Wireless network; Resource management
• Online Access: https://doi.org/10.1186/s13638-019-1616-9

Abstract In this paper, a novel resource management framework is introduced and exploited to ensure the efficient and smooth operation of a wireless network, assisted by an unmanned aerial vehicle (UAV), operating under the non-orthogonal multiple access (NOMA) scheme and consisting of both normal and malicious risk-aware users. User devices are assumed capable of splitting their transmission power in two different communication alternatives, established via either the UAV or the macro base station (MBS). The bandwidth offered by the UAV is accessible by everyone, delivers potentially higher rate of return taking into account the enhanced communication channel gains owing to its proximity to the serving users, but is prone to failure due to its potential over-exploitation. Accordingly, the UAV’s bandwidth is considered as common pool of resources (CPR). In contrast, the MBS’s bandwidth is considered as a safe resource offering to the users a more limited but guaranteed level of service, due to the fact that though it has less available bandwidth it operates under a more controlled access scheme. The theory of the tragedy of the commons is used to capture the probability of failure of the CPR, while the prospect theory is adopted to study the normal and malicious users’ risk-aware behavior in the UAV-assisted network. A non-cooperative power control game among the users is formulated and solved, in order to determine the users’ power investment to the dual communication environment. The existence and uniqueness of a Pure Nash Equilibrium point is shown and a distributed algorithm is introduced to converge to the PNE point. This overall resource allocation framework is intelligently exploited as the vehicle to detect malicious user behavior and therefore protect the network from the undesired effects of such behaviors. The performance and inherent attributes of the proposed user-centric risk-aware operation framework, in terms of its capability to effectively utilize the available system and user resources (i.e., bandwidth and power), while succeeding in identifying potential abnormal or malicious user behaviors is assessed via modeling and simulation, under different operation scenarios.