Broadcasting in Cognitive Wireless Mesh Networks with Dynamic Channel Conditions Muhammad Fahad, Junaid Qadir, Adeel Baig School of Electrical Engineering & Computer Science National University of Sciences and Technology H-12 Islamabad, Pakistan {fahad.manzoor, junaid.qadir, adeel.baig}@seecs.edu.pk Abstract—Recently, a variant of the popular wireless mesh network (WMN) technology, in which nodes communicate over multi-hop wireless links, has been proposed to address the problem of wireless interference and the scarcity of wireless spectrum. Such networks, which we shall call cognitive-radio wireless mesh networks (CWMN), can dynamically adapt network parameters according to existing conditions of the wireless channel. Such networks can also spot spectrum holes (or white-spaces, which are licensed spectrum slots currently not being used) and utilize them without causing interference to the incumbent users. There are numerous broadcast-oriented applications in WMNs, and in CWMNs by extension, that are sensitive to end-to-end delay. The inherent susceptibility of wireless links to interference, and the time-varying nature of wireless link’s quality, makes the design of efficient broadcasting algorithms for WMNs especially challenging. While minimum- latency broadcasting algorithms have been proposed in literature, they do not take into account the time-varying nature of wireless links in which channels may become unavailable due to interference (or, appearance of an incumbent user). Therefore, using the existing algorithms in CWMNs would necessitate expensive reconstruction of the tree in case of significant change in wireless topology due to interference or appearance of primary user. In this work, we propose a healing algorithm that addresses the problem of time-varying quality of wireless links and avoids expensive reconstruction of the entire tree by fixing the broken tree in a localized and decentralized manner. Simulation results show that our local healing algorithm improves both the broadcast latency and connectivity ratio (in face of interference) as compared to the approach of global reconstruction of trees. Keywords-local healing; broadcast latency; interference; cognitive radio networks; adaptive routing I. INTRODUCTION Wireless mesh network (WMN) [1], in which nodes connect over a multi-hop wireless network, has recently become a popular broadband wireless access technology and is being used in diverse scenarios like broadband home networking, community networking etc [4] [5]. Over the passage of time usage of WMN has increased manifold. To fulfill the increased usage and to drive future wireless applications, we need efficient usage of the available limited radio spectrum. The spectrum allocation, historically, has been managed in a command-and-control based approach in which a central organization licenses portions of the spectrum for dedicated use to various users/ technologies. Currently, almost all of the RF frequency spectrum, with some exceptions, has been allocated to licensed technologies/ users. Recent studies have concluded that while almost all the spectrum has been allocated, the actual usage of the spectrum is quite low by the licensed users. Thus, while an unlicensed user may have data to send, it cannot access the spectrum even when the licensed user does not have data to send. [8]. A new network architecture is in consideration that allow the secondary user (user who are not allocated specific spectrum) to use the spectrum whenever it is free in order to improve the utilization efficiency of the spectrum. The driving force behind this architecture is the software-defined-radio (SDR) technology using which radios can adapt to existing network conditions and change network parameters. Such adaptive radios are referred to as cognitive radios. We call a network in which all nodes are equipped with cognitive radios as a cognitive wireless mesh network (CWMN). Nodes in CWMNs use the spectrum opportunistically to transmit using wireless medium. For CWMNs, we also have to account for another form of interference, appearance of a primary user which wants to access the currently used spectrum which will render the channel unusable for the secondary user. Broadcasting in wireless medium is in general more challenging compared to wired medium due to the susceptibility of wireless medium to interference. In addition, the broadcasting problem in CWMNs with dynamic channel conditions is made more challenging by time varying wireless channel conditions that includes due to factors such as fading, multipath as well as activities of the primary user (PU). These two problems (interference and time-varying nature of wireless channels) make broadcast more challenging in the CWMNs where the channel assignment is being done dynamically depending upon the opportunities in the spectrum. If a broadcast tree is calculated for the CWMN then it is quite possible that the tree will not be valid over a significant amount of time. This maybe due to the dynamic nature of the network topology which is the result of time varying channel condition. Some of the links of the tree will not be connected due to the time varying channel condition. In this work, we are proposing an adaptive broadcast healing algorithm for use in CWMNs [2]. Our distributed and localized healing algorithm can adapt any of the currently proposed centralized or distributed broadcast tree construction algorithms (that construct a static broadcasting tree that does