A cognitive radio-based fully blind multihop rendezvous protocol for unknown environments

Abstract

In Cognitive Radio networking, the blind rendezvous problem is when two or more nodes must establish a link, but where they have no predefined schedule or common control channel for doing so. The problem becomes more challenging when the information about the existence of other nodes in the network, their topology, and primary user activity are also unknown, identified here as a fully blind rendezvous problem. In this paper, a novel and fully blind multihop (FBM) rendezvous framework is proposed with an extended modular clock algorithm (EMCA). The EMCA-FBM is a fully blind multihop rendezvous protocol as it assumes the number of nodes, primary radio activity and topology information as unknown. It is shown to work with different Cognitive Radio operating policies to achieve adaptiveness towards the unknown primary radio activity, and self-organization for autonomously handling the rendezvous process by using transmission schedules. It is capable of working without any information of neighbor nodes and terminating the rendezvous process whenever all or sufficient nodes are discovered. The proposed FBM is also shown to work as a general framework to extend existing single hop rendezvous protocols to work as a multihop rendezvous protocol. In comparison with other modified blind rendezvous strategies for multihop network, the combination of the proposed EMCA-FBM protocol and operating policies is shown to be effective in improving the average time to rendezvous (up to 70%) and neighbor discovery accuracy (almost 100%) while reducing harmful interference.

Introduction

Disasters are catastrophic events which cause great damage and require immediate attention. In the aftermath of a disaster, the first responders need to coordinate and victims need to communicate. However, the existing communication network infrastructure might be destroyed completely or partially which makes it difficult to proceed with the response efforts. New and rapidly deployable systems are needed to efficiently coordinate the response efforts and to provide communication services to the victims. Such a system can be deployed to replace a destroyed cellular base station temporarily or can be used independently to form a mesh network to provide services in a larger affected area.

In the early hours of a disaster, little may be known about the damage to existing communication systems, the radio environment and physical access, which makes it an unknown environment. A disaster response network is a network which can be deployed in the aftermath of a disaster to provide a response in the form of a temporary network, to replace an existing communication system, or connect two disconnected base stations [1]. For such unknown environments, a cognitive radio (CR) has the potential to provide an efficient and rapidly deployable network by sensing the radio environment, learning from the observed channel information and making decisions about the spectrum usage [1]. Spectrum information is sometimes provided through spectrum databases.

However, in disaster scenarios, the information in these databases is likely to be unreliable and incomplete, or the databases themselves may be unreachable. Therefore, a CR must dynamically sense the available radio channels and use them to establish communication if they are not occupied by a primary radio (PR). A PR in a disaster scenario can be an existing partially destroyed cellular network which still operates on some licensed frequency bands, a nearby radar station or TV station still operational, or some fixed bands for an emergency purpose like an ambulance or public safety service. Fig. 1, shows a disaster scenario for a cognitive radio network (CRN) including the cognitive or secondary and primary users. In this figure, different CR mobile base stations are shown to cover the affected areas where the existing communication networks are destroyed. These base stations can communicate with each other in a multihop manner to provide services like voice or data communication. The first responders can then use the CRs to extend the coverage of the service. In any case, the unknown PR activity makes it hard to find a common available channel to establish a rendezvous among nodes.

We define the Rendezvous process for a CRN as the completion of a handshake mechanism between two secondary radios on a single channel, which assumes that the two radios are within transmission range of each other, that they coincide on the channel for a sufficient time period, and that the channel has no detectable PR activity or excessive interference for the radios over that time period. Achieving a rendezvous in an unknown environment is a difficult task when nodes are not aware of the channel access sequence of the other nodes and when there is no pre-defined schedule or common control channel (CCC) available. A CCC is a pre-defined channel known by all nodes to exchange control packets among the network devices to establish communication. However, such CCC can be congested at times when the traffic or node density is high and negotiating a new CCC incurs excessive delays when PR activity is high [2], [3]. Channel hopping (CH) protocols can be used to avoid the use of CCC. Rendezvous can be achieved easily when two nodes follow the same CH sequence and are synchronized with each other in terms of time. The problem arises when nodes are not aware of the CH sequences of the other nodes and their wakeup times. Such problem is known as a blind rendezvous problem [3], [4]. The ‘blind’ element in that work has been limited to time synchronisation and to unknown channel information. In unknown environments like disasters, the nodes might also be unaware of the other information like the existence of other nodes, their topologies, their wakeup times and PR activity, which is identified here as a fully blind rendezvous problem. It introduces challenges like efficient rendezvous process termination, reliable neighbor information gathering, and synchronization to establish other network services with minimum network set up delay.

Existing blind rendezvous protocols do not provide an adaptive and self-organized multihop rendezvous protocol. Mostly, PR activity is not considered, and when considered the rendezvous delay is high. Although the existing protocols claim a guaranteed rendezvous within a bounded time, their guarantee remains valid only when the nodes choose different rate values to jump across the CH sequence to select a channel and when the PR remains inactive on a selected channel. For an efficient use of licensed spectrum, the standard bodies have already defined certain operating policies for a CR (IEEE 802.22 [5]). These operating policies suggest to not only vacate a channel on which PR is detected but also to avoid its use for some recommended time. However, the solutions so far do not consider such operating policies and their impact on the performance of a cognitive radio network is still unknown. Further, the need to start again the rendezvous process arises frequently when a PR appears frequently on an available channel. Such a situation of restarting the rendezvous process can be avoided by exchanging schedules for the future meeting point. In addition, a rendezvous process can be terminated easily when a node knows in advance the total number of nodes to discover. It is otherwise challenging when the nodes are not aware initially of the existence of the total number of nodes.

To address these challenges, our main contributions are summarised as follows:

• A fully blind multihop (FBM) rendezvous framework is proposed for multihop cognitive radio networks.

• The proposed framework is also presented as a general framework to enhance the functionality of existing single hop rendezvous protocol to multihop protocol.

• An extended modular clock algorithm with FBM (EMCA-FBM) is proposed for multihop network as a fully blind rendezvous protocol which assumes the number of nodes, PR activity, and topology information, as completely unknown.

• EMCA-FBM is shown to work with cognitive radio operating policies to achieve adaptiveness towards unknown PR activity.

• A termination strategy is proposed for unknown number of nodes, to terminate the rendezvous process when all or a sufficient number of nodes are discovered.

• An information exchange mechanism is presented to disseminate a complete network view among all nodes to autonomously handle the rendezvous process and to establish other network services.

The proposed EMCA-FBM is an extension of the work presented in [3], in which only single hop networks with a known number of nodes were considered. In this work an autonomous multihop rendezvous protocol is proposed to handle an unknown number of nodes. The proposed multihop protocol is shown to be adaptive towards the unknown PR activity; self-organized to facilitate new nodes entering or leaving the network; autonomous in handling the rendezvous process; and reliable in gathering the neighbor information. It is shown that the proposed multihop protocol together with cognitive radio operating policies can improve the time to rendezvous by up to 70% and can also achieve almost 100% neighbor discovery accuracy with a reduction in the average number of harmful incidents, in comparison with existing rendezvous strategies. Two operating policies, reactive and proactive, are also shown to be better at improving the rendezvous performance when compared with the basic Listen before Talk approach. For synchronization among the nodes, a reachability factor is shown also to be 100%, i.e., a message can successfully be forwarded to all the discovered nodes.

The rest of the paper is organized as follows. Section 2 discusses the related work of blind rendezvous protocols. The proposed fully blind rendezvous protocol is presented in Section 3 with system model. Section 4 discusses different CR operating policies. In Section 5, the simulation environment is discussed. Section 6 presents the performance evaluation. Finally, in Section 7 the paper concludes.