Joint Optimal Sensor Selection and Scheduling in Dynamic Spectrum Access Networks

Abstract

Abstract-Spectrum sensing is key to the realization of dynamic spectrum access. To protect primary users' communications from the interference caused by secondary users, spectrum sensing must meet the strict detectability requirements set by regulatory bodies, such as the FCC. Such strict detection requirements, however, can hardly be achieved using PHY-layer sensing techniques alone with one-time sensing by only a single sensor. In this paper, we jointly exploit two MAC-layer sensing methods-cooperative sensing and sensing scheduling-to improve spectrum sensing performance, while incurring minimum sensing overhead. While these sensing methods have been studied individually, little has been done on their combinations and the resulting benefits. Specifically, we propose to construct a profile of the primary signal's RSSs and design a simple, yet near-optimal, incumbent detection rule. Based on this constructed RSS profile, we develop an algorithm to find 1) an optimal set of sensors; 2) an optimal point at which to stop scheduling additional sensing; and 3) an optimal sensing duration for one-time sensing, so as to make a tradeoff between detection performance and sensing overhead. Our evaluation results show that the proposed sensing algorithms reduce the sensing overhead by up to 65 percent, while meeting the requirements of both false-alarm and misdetection probabilities of less than 0.01.

Improving the lifetime of multichannel cognitive radio sensor networks via new spectrum sensing method

“…In order to reduce energy consumption for CSS, some research works suggest that not all sensors participate in CSS . However, not only the number of sensors engaged in CSS but also the method of sensor selection affects the network lifetime.…”

“…Sensors should periodically sense all channels at the beginning of each frame to become aware of the presence of PUs. In order to make reliable decisions regarding the presence of PU, CSS is performed by a number of sensors at the start of each frame over each channel . For simplicity, we assume that OR rule is used in FC for the combination of received decisions about the presence of PU on each channel …”

“…Note that it is not necessary for all sensors to participate in CSS over an arbitrary channel . This is because the desired detection probability may be satisfied by the participation of only a number of sensors.…”

“…Therefore, a SU taking part in CSS consumes a nonnegligible amount of energy and time for CSS and consequently encounters more rapid drain of battery compared to those SUs not participating in CSS. In order to overcome the mentioned challenges, a number of research works have focused on the reduction of energy amount specifically consumed for CSS . In addition, in order to provide a green platform for CRNs, some of the existing researches proposed methods based on energy harvesting to enhance the battery lifetimes of CRNs .…”

Primary and secondary identification using energy detection with statistical primary information

“…power spectrum density. According to [5], for large N, test statistic can be approximated to Gaussian distribution as,…”