Article Preview
TopIntroduction
Over the last few years, the number of wireless devices has increased by multiple folds resulting in a spectrum scarcity problem. Studies anticipate that the situation will be further intensified by the increasing number of mobile subscribers that is expected to increase by almost a billion in 2020 compared to 2015 (“The mobile economy”, 2016) as well as the prediction that the number of connected devices will reach almost 13 billion by 2020 (Van der Meulen, 2015). Fortunately, studies and experiments, (Agarwal, Sengar, Gangopadhyay, & Debnath, 2016; Babalola et al., 2015; Dzulkifli, Kamarudin, & Rahman, 2011), have shown that the licensed users, also known as primary users (PU), are infrequently using their spectrum in certain locations at different times. Dynamic spectrum access (DSA) (Mitola, & Maguire, 1999) offered by the cognitive radio (CR) technology allows unlicensed users, called secondary users (SUs), to access the licensed radio spectrum when the licensed PUs are not using it.
To achieve the goals of the CR technology, and according to FCC (“Federal Communications Commission”, 2003), CR devices must have the cognitive capability of changing its transceiver parameters as well as the capability of making spectrum decisions that enable efficient spectrum utilization based on their analysis and learning of the spectrum. Similar to wireless networks, a group of communicating CR devices forms a cognitive radio network (CRN) that can be classified as either centralized or decentralized networks.
While CRNs can help in efficient utilization of radio resources among CR devices, unfortunately, CRNs introduce some challenges to the concept and goals of the CR technology. A major challenge among those facing the efficient operation of CRNs is efficient management of the available licensed spectrum among SUs with the aim of increasing the CRN throughput as well as increasing the spectral utilization while reducing the interference to other PUs and SUs at a guaranteed fairness among the CRN nodes.
Dynamic spectrum scheduling and management is composed of three main components: spectrum decision, spectrum sharing, and spectrum reallocation. Spectrum decision represents the ability of SUs (or the scheduler) to decide on the available spectrum that best suits the needed quality-of-service (QoS) requirements. Spectrum sharing represents the cooperation among the SUs to access a set of available/idle channels. On the other hand, spectrum reallocation is the ability of SUs to leave the licensed spectrum whenever a PU is detected and for the scheduler to guarantee efficient allocation of available channels.
This paper proposes an opportunistic switching-based and delay-aware scheduler that maximizes the total throughput for centralized CRNs. In contrast to conventional schedulers, where scheduling happens on slot-by-slot basis, the proposed scheduler is an inter-frame scheduler that aims at reducing the effect of hardware switching delay by performing the scheduling and allocation for multiple consecutive time slots. Additionally, the proposed scheduler does not only take into account the effect of spectrum switching delay but jointly considers the delay associated with the scheduling process, as well. Specifically, an optimization problem is formulated to find the optimum scheduler while taking into account the switching delay, the PU activity, historical information on the PU behavior, the channel quality, and the status of the SUs. The rationale behind scheduling the available PU channels to SUs over multiple time slots, instead of scheduling on slot-by-slot basis, is a try to reduce the unavoidable overhead of switching time. This is also done by estimating the throughput of every SU at each channel during these time slots.
The rest of this paper is organized as follow. The following section briefly discusses the problem while the following sections describe the system model and assumptions as well as the optimization problem. The performance and simulation results are demonstrated afterwards before the paper is summarized in the last section.