Distributed Resource Allocation for Cognitive Radio Ad-hoc Networks with Spectrum-Sharing Constraints Duy Trong Ngo and Tho Le-Ngoc Department of Electrical and Computer Engineering, McGill University, Montreal, QC, CANADA H3A 2A7. Email: duy.ngo@mail.mcgill.ca, tho.le-ngoc@mcgill.ca Abstract—In cognitive radio settings with highly dynamic pri- mary activities and with small opportunities for secondary access, the requirement to fairly distribute the temporarily available spec- tral ranges among the unlicensed users turns out to be of particular relevance. The current paper addresses this issue by presenting a new design formulation that aims to optimize the performance of an orthogonal-frequency-division-multiple-access (OFDMA) ad- hoc cognitive radio network, by means of joint subcarrier assign- ment and power allocation. Besides important constraint on the tolerable interference induced to primary network, to efficiently implement spectrum-sharing fairness, the optimization problem considered here strictly enforces upper and lower bounds on the total amount of temporarily available bandwidth to be granted to individual secondary users. Specifically, the system throughput is maximized via the application of Lagrangian duality theory. More importantly, the dual decomposition framework also gives rise to the realization of distributed solution. As the proposed distributed protocol requires very limited cooperation among the participating network elements, it is especially applicable for the ad-hoc networking environment under investigation, to which any central processing or control is certainly inaccessible. While the computational complexity of the devised algorithm is affordable, its performance in practical scenarios also attains the actual global optimum. The potential of the proposed approach is verified through asymptotic complexity analysis and via numerical examples. I. I NTRODUCTION To better utilize the radio spectrum, cognitive radio [1], [2] has been identified as an efficient technology to exploit the existence of the spectrum portions unoccupied by the primary (or licensed) users. While the primary users (PUs) still have priority access to the spectrum, secondary (or unlicensed or cognitive) users are permitted to have restricted access, subject to a constrained degradation on the PUs’ performance [3]. In spectrum sharing environments, the key design challenges of a cognitive radio network are therefore to guarantee a protection of the PUs from excessive interference induced by the secondary users (SUs) as well as to meet some Quality-of- Service (QoS) requirements for the latter. On the other hand, spectrum pooling is an opportunistic spectrum access approach that enables public access to the already licensed frequency bands [4]. The basic idea is to merge spectral ranges from different spectrum owners into a common pool, from which the SUs may temporarily rent spectral re- sources during idle periods of the PUs. In effect, the licensed system does not need to be changed while the SUs access unused resources. Orthogonal frequency division multiplexing (OFDM) has been recognized as a highly promising candidate for unlicensed users’ transmission in spectrum-pooling radio systems, due to its great flexibility in dynamically allocating the unused spectrum among the SUs as well as its ability to monitor the spectral activities of the PUs at no extra cost. In [5], the authors present a solution to an energy-efficient resource allocation problem that maximizes the cognitive radio link capacity, taking into account the availability of the OFDM subcarriers and the limits on total interference generated to the PUs. Based on a risk-return model, a convex optimization problem is formulated, which incorporates a linear average rate loss function in the objective function to include the effect of subcarrier availability. Considering networks with the coexistence of multiple primary and secondary links through OFDMA-based air-interface, reference [6] utilizes the dual framework from [7] to provide centralized and distributed al- gorithms that improve the total sum rate of secondary networks subject to interference constraints specified at PUs’ receivers. The work in [8] studies the optimization of an ad hoc cognitive radio network coexisting with multicell primary radio networks. To jointly optimize the throughput of the ad hoc SU links, Lagrange optimization is utilized to design fast-convergent sum rate maximization schemes constrained on the power spectral mask, the transmit power of SUs, the maximum-subchannel- rate, and the minimum-rate per SU link. Different from the existing approaches, we formulate in this paper a new design problem for OFDMA-based secondary ad- hoc networks to enhance the system throughput. In cognitive radio settings where the primary activities on the radio spectrum are highly dynamic and chances for secondary access are slim, the problem of fairly sharing the temporarily available frequency bands among the SUs becomes even more relevant than merely maximizing the system performance. Hence, in addition to the constraints on the tolerable interference limits induced by secondary network to the PUs and the maximum total transmit power at individual SUs, our formulation also incorporates the upper and lower bounds on the number of OFDM subchannels that unlicensed users are allowed to utilize. While the upper limits prevent cognitive users with favorable conditions from greedily filling all the spectrum holes, the lower thresholds provide certain guarantee of fairness in terms of bandwidth sharing to other SUs. Moreover, the dual optimization approach proposed in this paper allows to provide practically global optimal solution with affordable complexity, as opposed to highly demanding computational burden typically required by direct methods. Further, the dual design framework gives rise to the realization of distributed algorithm, which is certainly desirable for ad- hoc networks without any central coordination at all. In imple- menting the distributed scheme, we also introduce the concept 978-1-4244-5637-6/10/$26.00 ©2010 IEEE This full text paper was peer reviewed at the direction of IEEE Communications Society subject matter experts for publication in the IEEE Globecom 2010 proceedings.