CoRoute: A New Cognitive Anypath Vehicular Routing Protocol Wooseong Kim * , Soon Y. Oh * , Mario Gerla * , Kevin C. Lee † * Department of Computer Science, University of California, Los Angeles, USA Email: {wooseong, soonoh, gerla@cs.ucla.edu} † Cisco systems, USA Email: kevinl2@cisco.com Abstract—Vehicular communications promise to bring us safer driving and better traffic control. Dedicated Short Range Com- munications (DSRC) and IEEE 802.11p are now well established standards for the inter-vehicle and vehicle-to-road side unit (RSU) communication. These channels, however, are of limited capacity and are not sufficient to support the broad range of services envisioned in VANETs. Thus, vehicles will utilize WiFi (802.11 a/b/g) and unlicensed ISM band to acquire more capacity. Unfortunately, the WiFi channels in urban area are already heav- ily subscribed by residential customers. In this paper, we propose CoVanet, a cognitive vehicular ad hoc network architecture that allows vehicles opportunistic access to WiFi channels. CoVanet is the first approach to use cognitive radios in a VANET. It differs from conventional cognitive radio strategies in that it uses unlicensed band and operates in an ad hoc, multihop mode. In CoVanet, network topology and channel environment change frequently due to high node mobility. The main contribution of this work is a Cognitive Ad hoc Vehicular Routing Protocol (CoRoute) that utilizes geographical location and sensed channel information. Simulation results demonstrate CoRoute efficiency and robustness to mobility and external interference. Index Terms—Vehicular Networks, routing protocol, Cognitive Networks I. I NTRODUCTION Vehicular ad hoc networks (VANETs) have attracted atten- tion in the support of safe driving, intelligent navigation, and emergency and entertainment applications. Dedicated short range communication (DSRC) [1] has been standardized and exploited in vehicular testbeds, but it has only limited number of channels and it is basically reserved to safety purposes. Thus, VANETs must use WiFi for non safety applications. Unfortunately, WiFi channels suffer from scarcity of avail- able spectrum due to heavy interference from residential users, as well as various wireless devices in the ISM bands (e.g. 2.4GHz or 5GHz). Cognitive Radios, proposed by J. Mitola [2], are one of the solutions for spectrum scarcity in wireless networks. The cognitive radios opportunistically utilize spectrum holes in licensed bands without interrupting the licensed users (e.g., primary nodes (PN)). Cognitive radios are typically used in centralized, base stations, such as in IEEE 802.22 WRAN (Wireless Regional Area Network)[3] standard. However, cognitive radio implementations in ad hoc mobile environments still face a challenge due to the complexity of primary user detection and spectrum access scheduling. In this paper, we introduce CoVanet, a new cognitive VANET model operated on unlicensed bands instead of li- censed bands. It exploits multiple channels to increase network capacity and scalability while allowing a cognitive node (CN) to coexist with a PN in the same channel. Say, in CoVanet, residential 802.11a/b/g access points near roadside are PNs and vehicles are CNs. Based on this environment, we develop Cognitive Ad hoc Vehicular Routing Protocol (CoRoute), an anypath vehicular routing protocol that exploits channel and geo-location information. Vehicles periodically sense multiple channels in order to estimate channel workload and share the sensed channel information with each other. Each vehi- cle selects its own channel based on the measured channel information. Existing routing protocols for cognitive radios allocate channels along the routing path using well established metrics such as shortest expected transmission time (ETT) [4][5][6]. However, they fail to establish an optimal path in rapidly changing channel condition and workload. Some proactive routing protocols (SAMER) [7] utilize overall network link- state information (including channel and spectrum conditions) to dynamically calculate alternate paths. However, link-state routing protocols produce too much overhead to follow rapidly changing channel condition. Geographic based routing proto- cols, such as GPSR [8], GPCR [9] and GPSRJ+ [10], are more suitable to support robust connectivity with relatively low overhead even in high vehicle mobility. They, however, do not account for spectrum limitations and for interference and conflicts. CoRoute is the first attempt to VANET routing with cognitive radios accounting for both high mobility and spectrum scarcity. The remainder of this work is structured as follows: Sec- tion II describes our CoVanet architecture along with spectrum sensing and channel assignment algorithms. Section III gives an overview on the design of CoRoute. And simulation re- sults about CoRoute performance are shown in Section IV. Section V reviews related works. The paper concludes in section VI. II. CoVanet :COGNITIVE VEHICULAR AD HOC NETWORKS This section overviews CoVanet, a novel multi-radio multi- channel cross-layer architecture based on principles of cogni- tive radio system. 978-1-4577-9538-2/11/$26.00 ©2011 IEEE 766