Adaptive optimal relay selection in cooperative vehicular communications under security constraints

Abstract

In this paper, we study the physical layer security of multiple decode-and-forward cooperative vehicular to vehicular communications in which a known eavesdropper can receive the signals from source and relay nodes in the broadcast and cooperative phases, respectively. At first, we present a generalized optimal relay selection (GORS) in order to select the best relay considering the security of both broadcast and cooperative phases. Then, we apply an incremental mechanism to this generalized version, and propose an adaptive optimal relay selection (AORS) scheme which adaptively provides and preserves the security. Two scenarios are considered in the network based on availability of channel state information (CSI) of the main and wiretap links. For the first scenario where CSI is available at the receiver nodes, we analyze the secrecy outage probability and average spectral efficiency. Whereas, for the second scenario where CSI can be available at the transmitter node of each phase, the average secrecy capacity is analyzed. The results show that the AORS scheme can improve the secrecy performance compared to the GORS and Direct schemes. Moreover, the asymptotic analysis is provided which shows that the proposed scheme provides full secrecy diversity order. Finally, results of Monte Carlo simulations are presented that denote validity of the analysis and evaluate the system performance.

Introduction

Cooperative communication has recently gained significant attention in the enhancement of security through physical layer characteristics. In this approach the relay nodes are applied between the source and main destination nodes in order to provide a higher capacity for the main link compared to the wiretap one [1], [2], [3]. It has also been shown that increasing the number of the relays enhances the secrecy performance [4]. Furthermore, relay selection (RS) is a well adopted scheme with the multiple relay scenarios which can improve the performance of the cooperative systems with or without security constraints [5], [6]. Depending on passivity or activity of the eavesdropper, various RS schemes can be applied. In detail, in the case of the passive eavesdropper, a relay is selected conventionally applying the CSI of the main link [6], while an optimal relay selection (ORS) can be implemented based on the availability of the CSI of the main and wiretap links in the case of a known eavesdropper [6], [7], [8], [9], [10].

Broadly speaking, the physical layer security of wireless relay networks has been commonly studied from the cooperative phase (CP) perspective [7], [8], [9], [10], [11], [12], [13], [14], [15], [16], [17], [18], [19], [20], [21]. However, in an operational environment where there are direct links from the source node to the main destination and eavesdropper nodes, the issue needs further investigation with additional emphasis on the broadcast phase (BP). It is remarked that the majority of studies such as [7], [8], [9], [10], [11], [12], [13], [14], [15], [16] have focused on the security of the second phase; they assumed that just the relays can receive the source signal which means that there is no wiretapping in the BP phase. Whereas, we know that the eavesdropper might wiretap from both source and relay nodes in operational environments. Different from those works, in [17], [18], [19] the authors investigated the general form of cooperative systems which contains the direct links from the source node to the main and eavesdropper nodes. However, they presented the secrecy capacity at the end of the CP phase without paying attention to being exposed the secret message in the BP phase. It should be noted that in the case of the half-duplex constraint which separates the CP and BP phases, the cooperation of the relays in the second phase is useless if the secret message gets revealed in the first phase. This implies that the overall secrecy capacity between the source and main destination be different with the secrecy capacity considered in [17], [18], [19]. Moreover, in [20] and [21], the authors assumed that the links between the source and relays experience a higher quality compared to the link between the source and eavesdropper. Clearly, by this assumption the BP phase does not limit the security, and thus one can consider the secrecy capacity at the end of the CP phase. Generally, it can be said there are few works such as [22] and [23] which have considered the security of the cooperative systems with wiretapping in both BP and CP phases according to our approach in this paper. In [22] which has established the utility of the cooperative strategies in the physical layer security, the information theory aspect of the issue was investigated for a four-terminal relay–eavesdropper channel and an outer-bound on the optimal rate-equivocation region was derived. In that work the authors proved theoretically the gain offered by the proposed cooperation strategies and validated it in the additive white Gaussian noise (AWGN) channel numerically. In [23], the authors investigated the role of the cooperative relays to provide and improve secure communication rates in a full-duplex multiple relay network. It should be noted that they correctly have calculated the secrecy capacity at the end of the CP phase because the BP and CP phases take place simultaneously due to the full-duplex assumption. However in the case of half-duplex their work is not applicable.

Here, we study and analyze the physical layer security of a multiple DF relay cooperative system in the presence of a known eavesdropper which overhears from both source and relay nodes. As an extension of the ORS strategies in [6], [7], [8], [9], [10], here we present the generalized optimal relay selection (GORS) in which the best relay is selected based on secrecy capacities in both the BP and CP. Then, we propose an incremental version of GORS named as adaptive optimal relay selection (AORS) with respect to the existence of the direct links and the availability of the CSI of the main and wiretap channels. Please note that the AORS scheme is different from the previous incremental and selection approaches which are only based on the CSI of main link [17], [18], [19], [20], [21]. Meanwhile, the GORS is different from ORS schemes presented for cooperative systems of no direct link in [6], [7], [8], [9], [10], [11], [12]. We derive analytical expressions for the secrecy outage probability and average spectral efficiency of the AORS scheme when the channel information is available at the receiver nodes. Whereas, the average secrecy capacity of the proposed scheme is analyzed for the case that the channel information is available at the transmitter nodes. In addition, we analyze the asymptotical behavior of the system and derive high signal-to-noise (SNR) expressions for the secrecy outage probability which show that the proposed AORS scheme provides $M+1$ secrecy diversity order when M relays are employed.

The rest of this paper is organized as follows: The system model and adaptive cooperative scheme are presented in Section 2. In Section 3, we analyze the secrecy performance of the proposed scheme. In Section 4, numerical results are presented, and finally the paper is concluded in Section 5.