1 Outage Performance of the Primary Service in Spectrum Sharing Networks M. G. Khoshkholgh, K. Navaie, Senior Member, IEEE, and H. Yanikomeroglu, Member, IEEE Abstract—In this paper, we utilize stochastic geometry to analyze the primary service outage performance for spectrum sharing in Rayleigh fading environment. Using this approach, the impacts of the secondary service parameters and wireless environment on the primary service (PS) outage probability are analyzed. We further obtain a closed-form for the PS outage probability. The maximum secondary service transmitter node density for a given outage probability constraint of the PS is then obtained. We also investigate the impact of secondary spectrum sensing on the PS outage probability. A novel approach is further proposed which provides tight approximation for the PS outage probability. The results of the proposed approach are then validated through analysis and simulations. We then consider power control in the secondary network and show that the truncated channel inversion power control significantly decreases the PS outage probability. Cases with centralized and decentralized cooperative spectrum sensing are also studied and their corresponding PS outage probabilities are analyzed. Mean spatial throughput of the secondary service (SS) is also analyzed. We further investigate the impact of the PS outage constraint on the spatial throughput of the SS. Extensive simulations confirm our analytical derivations. Index Terms—Outage probability, spectrum sensing, spectrum sharing, stochastic geometry. ✦ 1 I NTRODUCTION T O IMPROVE the utilization of the allocated frequency bands, spectrum sharing is proposed by the Federal Com- munications Commission (FCC) [1]. In this method, under certain conditions, a Secondary Service is able to access to a frequency band formally allocated to the Primary Service [2], [3]. Various schemes are proposed in the literature for spec- trum sharing (see, e.g., [4]) where they are usually referred to as the Dynamic Spectrum Access (DSA). In the DSA, the secondary service (SS) dynamically detects and utilizes the spectrum holes or white spaces [3], [5]. White spaces are those parts of the spectrum allocated to the pri- mary user which are under-utilized in some particular times and specific locations. Here, we focus on the overlay spec- trum sharing which is also referred to as the Opportunistic Spectrum Access (OSA) [4]. In the OSA, the SS utilizes spectrum sensing schemes to detect the white spaces. If the spectrum condition is detected as ‘idle’ (i.e., the primary service (PS) is inactive) the SS is allowed to transmit its own information, otherwise, (i.e., the primary service is active), the SS is not allowed to transmit. Therefore, given perfect spectrum sensing, adopting OSA increases the spectrum efficiency while imposes no negative impact on the PS. In practice however, spectrum sensing schemes are usually imperfect [3], [4]. The inaccuracy in the spectrum sensing, results in degrading Quality of Service (QoS) of the PS users. Furthermore, by increasing the number of the secondary service users, the spectrum sensing inaccuracy may seriously affect the performance of the primary network. Therefore, in the spectrum sharing, it is essential to manage the SS parameters such that the PS QoS constraints including the • M. G. Khoshkholgh and H. Yanikomeroglu are with the Department of Systems and Computer Engineering, Carleton University, Ottawa, ON, Canada. K. Navaie is with the School of Electronic and Electrical Engineering, University of Leeds, Leeds, UK. E-mail: keivan@sce.carleton.ca Part of this paper has been presented in the IEEE ICC’10, May 2010. outage probability and achievable capacity are always kept satisfied. One of the main parameters to be adjusted is the maximum number of the SS users in a given coverage area of the primary network. To obtain the maximum number of the SS users, in this paper, we employ stochastic geometry (see, e.g., [6]) which is shown to be a very powerful mathematical tool for performance evaluation of wireless networks (see, e.g., [7], [8], [9], [10], and [11]). A comprehensive tutorial on the stochastic geometry and its application in wireless communication networks can be found in [12] and [13]. In [14] based on homogenous Poisson Point Process (PPP), upper and lower bounds on the transmitter node density are obtained to satisfy the outage probability constraint in Code Division Multiple Access (CDMA) based ad hoc net- works. The impact of fading, power control, and interference cancelation on the outage probability in ad hoc networks are also studied in [15] and [16]. Stochastic geometry has been also successfully adopted for connectivity modeling in ad hoc networks [17]. In [18], [19], the authors utilize advanced stochastic geometry to design efficient hierarchical sensor networks with the objective of minimizing the network energy consumption. In the related literature, stochastic geometry is also em- ployed to analyze spectrum sharing performance. Chan- nel capacity of the SS is the focus of [20]. Furthermore, in [21] distance-dependence path-loss attenuation model is considered and log-normal distribution is proposed as an approximated probability distribution function (PDF) of the aggregate interference at the PS receiver. Considering fading effects of the wireless channel as well as spectrum sensing parameters, [22] proposes an approximation for the distri- bution of the interference aggregation at the PS receiver. They also analyze the impact of cooperative spectrum sensing among the SS users on the primary service outage probability. However, they assume that the SS transmitters are able to identify the spectrum status using signaling channels among the PS receiver, and the SS transmitters. Further their work Digital Object Indentifier 10.1109/TMC.2012.156 1536-1233/12/$31.00 © 2012 IEEE IEEE TRANSACTIONS ON MOBILE COMPUTING This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication.