Sensing and Transmitting Schedule in Energy Harvesting Powered Cognitive Radio Network with Awareness of Secrecy Capacity

Abstract

The proliferation of wireless communication technology has been raising the demand for more and more improvement in data transmission capacity under limited radio spectrum resource. Meantime, the static spectrum allocation policy of government agencies assigned to licensed users, or primary users (PUs), leads to inefficient utilization of a large amount of licensed spectrum. Therefore, the scarcity and under-utilization of radio spectrum resource have driven the concept of cognitive radio (CR). CR technology is a communication paradigm which allows non-licensed users, or cognitive users (CUs), to dynamically and opportunistically access spectrum holes (licensed spectrum bands are not being utilized at a particular time and a specific geographic location) that are temporally unoccupied by PUs. In addition, energy-harvesting powered CR networks (CRNs) have lately become attractive research issues in the literature. Although harvesting capacity has been limited and still need to be improved, energy-harvesting envisions to liberate the CUs in CRNs from energy constraints. Up to now, there have been more and more efforts in the literature to improve energy-harvesting capacity in the future. In the meantime, (i) limits in energy harvesting capacity should be considered as one of the most important criteria for the design of energy-harvesting powered-CRNs. In addition, similar to traditional wireless networks, (ii) CRNs have also some security vulnerabilities i.e. malicious attacking, eavesdropping which should be properly addressed in the design of these networks in future. Furthermore, the demand to improve the wireless communications rate under the scarcity of spectrum resources leads to (iii) the concept of full-duplex (FD) transmission protocol, which allows a radio device to simultaneously transmit and receive on the same frequency band. Recently, FD protocols represents as far as an attractive option for increasing the throughput of CR systems.

Generally, CRNs are not allowed to make any interference to PUs. Therefore, robust and reliable spectrum sensing (SS) schemes to detect PUs’ signal are utmost important for the operation of CRNs. To the best of our knowledge, SS schemes and transmitting power allocation algorithms for infinite-energy CRNs have been well-studied in literature up to now. However, sensing-transmitting schedule, transmission power allocation schedule, and transmission protocol switching schedule in energy-harvesting powered-CRNs with and without awareness of secrecy capacity are still under-investigated. Motivated from the foregoing survey, this dissertation will address these remaining challenges for energy-harvesting powered-CRNs as follows:

Firstly, considering channel switching delay and imperfect sensing, we proposed an optimal multi-slot multi-channel sensing order for the opportunistic access to a number of potential primary channels. In addition, the correlation of channel availability statistics across time slots and channels is also considered. The problem was formulated and solved based on the partially observable Markov decision process (POMDP) framework and the optimal stopping theorem. The goal of this work is finding an optimal sensing order of channels in order to maximize throughput of CU in CRNs.

Secondly, based on the theory of optimal stopping, we propose an algorithm to optimize the sequential cooperative SS and reporting process in which the fusion center (FC) sequentially asks each CU to report its sensing result until the stopping condition which provides maximum expected throughput for CRN is satisfied. Simulation shows that performance of the proposed scheme can be improved by further shortening the reporting overhead and reducing the probability of false alarm compared to other schemes in the literature.

Thirdly, considering energy-harvesting powered-CRNs utilizing multiple potential primary channels, we propose a scheme to find an optimal channel-sensing and transmitting schedule, consisting of finding (i) the optimal action (silent or active) and sensing order of channels and (ii) the optimal amount of transmission energy corresponding to the channels in the sensing order, for the operation of the CU in order to maximize the long-term expected throughput. The performance of the proposed scheme is evaluated in comparison to related schemes in the literature which only considered immediate throughput.

Next, we considered a practical scenario of energy-harvesting powered-CRNs under the presence of eavesdropper(s). A pair of CUs opportunistically accesses a potential licensed channel; meanwhile, they should ensure that their confidential communications are not leaked to the eavesdropper. Based on expected secrecy transmission rate calculated over subsequent $K$ time slots, we proposed a scheme to find an optimal spectrum sensing and transmitting schedule as well as the optimal amount of transmission energy in each processing time slot. In particular, the CUs decide either (i) to sense the channel and transmit its data if the channel is found vacant or (ii) to stay silent during the current time slot to save energy as well as to wait for more harvested energy for use in the next time slots. The proposed scheme aims to improve long-term secrecy transmission rate of CRNs.

Finally, we focuses on utilizing the full advantages of both half-duplex (HD) and FD transmission protocol for the operation of CU in energy-harvesting based-CRNs. Based on the available probability of potential primary channel, information about cognitive channel gain between cognitive base station (BS) and CU, and the amount of energy-harvesting rate, we propose a scheme to find an optimal switching schedule between HD and FD transmission protocol as well as the amount of transmission energy for the BS and CU corresponding to each transmission protocol for maximizing long-term expected transmission rate of the BS-CU transmission pair. Then, the problem is formulated and solved by adopting the POMDP-framework. Simulation shows that average throughput attained by the proposed scheme is greatly improved compared to that of the conventional scheme, and remarkably, when the energy-harvesting rate becomes low. We close our studies by raising another solution to this problem by adopting the Actor-Critic learning framework. Although the actor-critic learning process may converge to a locally optimal policy, this method generates the action directly from training policy; hence, it requires much less formulation and computation to select an action compared to POMDP framework.