Spectrum Allocation Framework for Multiuser Cognitive Radio Systems Mohammad Iqbal Bin Shahid and Joarder Kamruzzaman Faculty of Information Technology, Monash University, Australia. Email:{M.I.Shahid, Joarder@infotech.monash.edu.au} Abstract—One of the most challenging issues in cognitive radio networks is to dynamically access the radio frequency spectrum in an uninterrupted manner. To achieve this, omniscient allocation of spectrum bands among cognitive radio users is crucial. Most of the existing spectrum allocation methods select a band from a pool according to the service requirements of a single user, neglecting the demand of multiple users. In this paper, we introduce a collaborative framework for allocating multiple bands among multiple secondary users. The proposed method defines a capacity of service metric based on the optimal sensing parameters and utilizes this metric to assign distinct bands to all or highest possible number of contending users. Performance evaluation suggests that the proposed method exhibits significant superiority over conventional approaches in terms of improved throughput and spectrum utilization, reduced interference loss and collision, and hence, enhances dynamic spectrum access and sharing capabilities. I. I NTRODUCTION Recent interest in cognitive radio technology indicates a major shift in policies and approaches in using the radio frequency spectrum. Opportunistic spectrum access is now considered a better alternative to the typical fixed spectrum licensing policy [1] which is inefficient and insufficient to cope with new applications demanding more spectrum resources. Cognitive radio is the key technology in implementing intel- ligent and flexible networks to meet future spectrum demand where radio devices access the radio frequency (RF) spectrum in an opportunistic fashion. The operation of a cognitive radio (CR) in accessing the RF spectrum involves three major mechanisms: spectrum sensing, spectrum use, and spectrum switching. Spectrum sensing is accomplished by an unlicensed secondary user(s) by finding a white space (absence of licensed primary user) in licensed bands. The most widely accepted method for sensing white spaces is energy detection based cooperative spectrum sensing [2], [3]. After finding a suitable band, the secondary user starts using that band and whenever it senses the presence of the primary user (PU), it quickly evacuates the band to inflict minimum interference on the primary system [4]. Finally, the secondary user has to find and switch to a vacant band to resume transmission. Selecting a suitable spectrum band from a set of licensed bands is a critical issue in CR networks. This depends on the availability of spectrum bands, sensing accuracy regarding their occupancy, and quality of service requirements by the (sec- ondary) user. Considering these factors, the most suitable band is assigned to the requesting user. Band selection and allocation process becomes even more complicated if the number of users is more than one. In this case multiple bands have to be selected from the available vacant bands according to some measures and allocated among requesting users. Spectrum pooling [5] is proposed as a solution to this problem where a user can select one of the pooled channels for transmission if it is not occupied at that time. Selection of spectrum bands from a pool has been investigated extensively in the literature. Among them, some notable works related to CR networks are mentioned below. In [6], a spectrum selection method is proposed where spectrum sensing is performed only over a number of spectrum bands selected according to their sensing capacity. This method considers periodic sensing and the ratio of sensing and trans- mission durations is optimized to achieve maximum sensing efficiency. A game theoretic framework to analyze the behavior of cognitive radios for distributed adaptive channel allocation is proposed in [7]. It is shown that the channel allocation problem can be formulated as a potential game between two users, and thus converges to a Nash equilibrium point for deterministic channel allocation. A multiband joint detection framework for wideband spectrum sensing in individual cog- nitive radios is proposed in [8]. In this framework, a bank of multiple narrowband detectors is jointly optimized in order to improve the opportunistic throughput capacity of cognitive radios keeping the interference to the primary communication systems below a certain value. However, none of these methods consider multiple secondary users, where it is highly likely that two or more users may simultaneously request to access the RF spectrum, and in some cases, the same spectrum band. In addition, they consider periodic sensing and assume a priori knowledge on PU usage pattern which somewhat reduce spectrum utilization and generality. A method for accommodating multiple users into a number of spectrum bands is proposed in [9]. The impact of various factors on secondary user performance, namely, channel idle time and probability, channel correlation, and sensing schemes are considered to maximize the secondary user throughput. This method divides the target spectrum bands into a number of blocks and ‘squeezes’ a number of secondary users into these blocks, ignoring their band requirements. Consequently, aver- age throughput reduces when the number of users increases. Moreover, continuous rearrangement of spectrum bands already occupied by some secondary users costs a big overhead and lowers spectrum utilization. Furthermore, this method over- looks the effect of expected interference on the primary system when a band is considered for selection. 978-1-61284-231-8/11/$26.00 ©2011 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE ICC 2011 proceedings