Coordination problem in cognitive wireless mesh networks

https://doi.org/10.1016/j.pmcj.2012.09.001Get rights and content

Abstract

Due to their potential to create and extend pervasive communication applications to cognitive environments with distributed control, the emerging technology of cognitive wireless mesh networks is gaining significant attention from a growing research community. However, the major challenge in cognitive networks is the adaptation to time and space variability of the available resources, namely chunks of the frequency spectrum called channels. In particular, this problem is exacerbated in cognitive mesh networks because there exists no direct communication among devices which thereby cannot establish a global (common) control channel to coordinate the entire network. Instead, only local control channels that vary depending on the time instant and location, can be established to coordinate cognitive devices among themselves. This paper first analyzes the underlying challenges and existing approaches to address the absence of a static and global control channel, and then propose a novel Control channel formation protocol, called Connor. Our protocol Connor is a fully distributed coordination scheme where cognitive mesh devices self-organize into clusters based on the similarity of available channels and on topological constraints. Compared with the existing clustering algorithms, which requires synchronization, the proposed Connor performs better in most cases without imposing synchronization.

Introduction

An emerging technology in wireless environments is the concept of cognitive networks (CNs) that can handle situations in which the licensed frequency spectrum is underutilized while the unlicensed spectrum is overcrowded. Being easily maintainable, CNs are continuously improved and upgraded in a way that is completely integrated with the surrounding environment [1]. Hence, CNs are particularly efficient when a self-organizing network is desired. However, the challenge is how to deal with the time and space variability of available resources, namely the frequency spectrum. Indeed the major challenges in CNs are [2]: (i) sensing the radio environment to detect spectrum holes in terms of both time and location, (ii) controlling an efficient employment of the spectrum holes, and (iii) allocating power in each spectrum hole.

Fig. 1 shows a CN with Primary Users (PUs) which are licensed users, such as Digital TV transmitters (DTVs), microphones, and other devices that require a license and payment to use the frequency spectrum. A CN also consists of Secondary Users (SUs) that are unlicensed devices intending to opportunistically grab unused channels without causing any harmful interference to the Primary Users (PUs). Clearly, SUs need a cognition of the wireless environment and hence are called cognitive devices. All the complexity due to spectrum sharing is borne by the cognitive devices, thus deploying rules to opportunistically operate in the licensed spectrum and being essentially invisible to the PUs. Therefore the PUs do not require any change in their spectrum management implying the cognitive devices are capable of dynamic spectrum use.

Such cognitive devices can be used to design a wireless mesh network, called a cognitive wireless mesh network (CWMN). A CWMN considers point-to-point wireless links where connections are established directly between the network devices, thus implementing a decentralized model. In other words, a pre-existing infrastructure, such as the access points in a managed wireless network, does not exist. The basic difference with access networks is that mesh networks do not provide direct access to the Internet; instead each device can obtain an Internet connectivity through multiple hops. A CWMN houses mesh routers and mesh clients such that the mesh routers create what is called a backbone, a multi-hop network interconnecting mesh routers. A particular type of mesh routers can be identified in the backbone, called mesh gateways, that support Internet connectivity. Each mesh client is connected to a backbone device in order to have its packets forwarded from/to the Internet; hence the backbone forms a multi-hop route between mesh clients and the Internet. Hereafter we focus only on the network backbone as the main component to provide the Internet connectivity. We also assume that cognitive mesh devices (CMDs) are SUs belonging to the network backbone, and that they sense the environment, learn from it and avoid harmful interference to the PUs.

CWMNs are attracting attention from a growing community of researchers due to their potential and ability to create and extend self-organizing networks without requiring infrastructure support. However, CWMNs pose significant challenges, such as resource discovery, topology changes and resource allocation [3]. In CWMNs, the presence of PUs implies that available resources vary in each device in time and space, and are thus local instead of being identical in all the devices as assumed in the traditional WMNs. Hence, CMDs have local knowledge of available resources but in order to establish end-to-end paths in a mesh environment, they need to know what resources are available at least at their one-hop neighbors and possibly beyond such devices. For this reason, information on resource availability must be disseminated to ensure point-to-point and consequently end-to-end communications. In the literature [4], this issue is referred to as the coordination problem, which is identified as the main challenge in CWMNs because, in such networks, no Internet connectivity is guaranteed yet the coordination needs to be established on the wireless links.

In order to address the coordination problem, devices need to communicate on a common channel, which is a chunk of the frequency spectrum. However, the channel cannot be static and globally controlled due to the variability of the available frequency spectrum. Hence, network partitioning approaches have been proposed in the literature. The most common partitioning technique is the clustering approach where linked devices are partitioned into groups sharing common interests and characteristics. This paper starts with a detailed description of the coordination problem and clustering approaches, and then presents a novel protocol, called a Control channel formation protocol (Connor).

Connor is a fully distributed coordination scheme for CWMNs where devices self-organize themselves into clusters, based on the similarity of available channels and topological constraints. In each cluster, cognitive devices share a common control channel and exchange their individual sensing results which are combined to obtain final results. The characteristics of Connor are: (i) no synchronization among cognitive mesh devices is required; (ii) coordination among cognitive mesh devices is assured; (iii) fast and efficient adaptability to changes in the surrounding environment, which is a fundamental property in CWMNs. Finally, we show that Connor efficiently forms a limited number of clusters with common control and backup channels. Our experimental results demonstrate that Connor performs better in most cases compared with the existing clustering algorithms, which impose synchronization.

The paper is organized as follows. In Section 2, we analyze the issues related to implementing CWMNs, in particular focusing on the coordination problem. Section 3 reviews the relevant literature concerning common control channel approaches for cognitive networks while Section 4 focuses on clustering as the most promising technique for CWMNs to address the absence of a static and global coordination mechanism. Section 5 lists different types of control channels that can be used whereas Section 6 describes and analyzes the proposed Connor protocol. Section 7 presents the performance of Connor and its comparison with an existing protocol SyncCFP [4]. Section 8 concludes the paper along with directions for future research.

Section snippets

CWMN issues

Topology changes affect WMNs and CWMNs because routes formed over multiple hops may periodically experience disconnections. This condition is further exasperated in CWMNs due to the presence of PUs. In fact, from device to device, the available resources vary in time and space, and are also local instead of being identical in all the CMDs. This makes the establishment of end-to-end paths harder because a CMD does not know what views the other CMDs have about the environment. That is why

Related works

In recent years, several studies have been reported on the design and analysis of CWMNs. A comprehensive overview on challenges and early solutions was presented in [6], which focused on the most promising approaches and commented on the future research roadmaps.

Although several approaches exist for tackling the control channel problem in WMNs, they are not applicable to CWMNs due to the time and space variability of available resources, i.e., wireless channels. For example in WMNs, network

Clustering approaches

The control channel problem addresses the necessity of network devices to coordinate among themselves to satisfy network requirements and perform resource allocation algorithms. Moreover, they need coordination mechanisms to perform spectrum access, network topology control, transmission power control, and bandwidth allocation. In general, the required information for each network device can vary depending on the algorithm they run. In a CWMN, cognitive mesh devices (CMDs) want to exchange

Types of control channels

The control channel problem has been addressed in the literature with several approaches. In the following we first distinguish static and dynamic approaches, followed by in-band and out-of-band control channel approaches.

Connor: control channel formation protocol

We proposed Connor as a fully distributed cluster-first control channel formation protocol that aims at disseminating information and addresses the coordination problem, hence the control channel problem in CWMNs. Using Connor, cognitive mesh devices self-organize themselves into clusters, based on the similarity of unused channels and on topological constraints. To the best of our knowledge, there exists no technique for the control channel problem in CWMNs using fully distributed clustering.

Performance evaluation

The experiments were conducted using an event-driven simulator for cognitive networks that we developed and is available for download at [26]. The simulator is in C++ and it is derived from NS-2. Our simulator models CNs and measures iterations between cognitive devices. The implemented MAC (medium access control) layer takes the IEEE 802.22 standard as reference where the frame is 10 ms and consists of a downlink sub-frame, uplink sub-frame and Coexistence Beacon Protocol (CBP) burst. The

Conclusions

This paper has dealt with cognitive wireless mesh networks (CMWNs) and in particular, the coordination problem, in an environment characterized by time and space variability of available resources. After analyzing challenges and approaches proposed in the literature for the coordination problem. We proposed a novel clustering algorithm, called Connor for the control channel problem. Connor does not require synchronization among cognitive mesh devices and allows rapid re-clustering when changes

Acknowledgments

The authors would like to thank the anonymous reviewers for their valuable comments and suggestions that helped improve the quality of the paper.

References (26)

  • B.F. Lo

    A survey of common control channel design in cognitive radio networks

    Elsevier Physical Communication

    (2011)
  • V. Gardellin et al.

    G-PaMeLA: a divide-and-conquer approach for joint channel assignment and routing in multi-radio multi-channel wireless mesh networks

    Journal of Parallel and Distributed Computing

    (2011)
  • K.J.R. Liu

    Advances in cognitive radio networks: a survey

    IEEE Journal of Selected Topics in Signal Processing

    (2011)
  • V. Gardellin, S.K. Das, L. Lenzini, Cooperative vs. non-cooperative: self-coexistence among selfish cognitive devices,...
  • M. Cesana et al.

    Networking over multi-Hop cognitive networks

    IEEE Network

    (2009)
  • J. Zhao et al.

    Spectrum sharing through distributed coordination in dynamic spectrum access networks

    Wireless Communications & Mobile Computing

    (2007)
  • I.F. Akyildiz et al.

    Spectrum management in cognitive radio ad hoc networks

    IEEE Network: The Magazine of Global Internetworking — Special issue title on networking over multi-Hop cognitive networks

    (2009)
  • H.A.B. Salameh et al.

    Channel access protocols for multihop opportunistic networks: challenges and recent developments

    IEEE Network

    (2009)
  • Y. Minghao, H. Lianfen, C. Huihuang, Multi-channel MAC protocol for cognitive wireless mesh network, in: IEEE...
  • Y. Song et al.

    Joint channel and power allocation in wireless mesh networks: a game theoretical perspective

    IEEE Journal on Selected Areas in Communications

    (2008)
  • S. Sengupta, M. Chatterjee, R. Chandramouli, A coordinated distributed scheme for cognitive radio based IEEE 802.22...
  • T. Chen, H. Zhang, M.D. Katz, Z. Zhou, Swarm intelligence based dynamic control channel assignment in cogmesh, in: IEEE...
  • L. Lazos, S. Liu, M. Krunz, Spectrum opportunity-based control channel assignment in cognitive radio networks, in: IEEE...
  • Cited by (14)

    • Adaptive variable-size virtual clustering for control channel assignment in dynamic access networks: Design and simulations

      2021, Simulation Modelling Practice and Theory
      Citation Excerpt :

      Many distributed coordination mechanisms have been presented to organize the CR users into groups based on the fact that CR users located in the same geographical area typically share same spectrum opportunities. In such schemes CRN divided into groups called cluster, each cluster members select local CCC to exchange control packets [17–22]. In [17], the authors proposed a cluster based CogMesh scheme to form one-hop neighboring clusters, each cluster has a cluster head (CH) that selects CCC and control cluster operations.

    • Spread spectrum-based coordination design for spectrum-agile wireless ad hoc networks

      2015, Journal of Network and Computer Applications
      Citation Excerpt :

      The use of two different types of underlying CCs were proposed in Gardellin et al. (2013), one for global coordination and the other for local coordination. The proposed design in Gardellin et al. (2013) employs code division multiple access (CDMA) with adaptive frequency hopping in order to protect PR users and achieve reliable CR communications. It is worth noting that most of the previously proposed underlay CC designs (e.g., Wasden et al., 2012; Gardellin et al., 2013) provide only design frameworks based on underlay spectrum access strategies, with no insights into CC protocol design, implementation and performance in a multi-user scenarios.

    • A Leader Election Protocol for Cognitive Radio Networks

      2017, Wireless Personal Communications
    View all citing articles on Scopus
    View full text