Load aware traffic engineering for mesh networks
Introduction
Wireless Mesh Networks (WMNs) [1], [2], [3] have emerged recently as a promising technology for next-generation wireless networking to provide better services. A WMN consists of two types of nodes: mesh routers and mesh clients. Mesh routers form the backbone and they have minimal mobility which guarantees high connectivity, robustness, etc. The mesh client nodes can be stationary or mobile. A simple example of Wireless Mesh Network is presented in Fig. 1.
Like ad hoc networks, each node operates not only as host but also as router, forwarding packets to and from an Internet-connected gateway in a multi-hop fashion. Wireless Mesh Networks are considered as a type of ad hoc networks. But, because the aim of WMN is to diversify the capabilities of the ad hoc network, more sophisticated algorithms and design principles are required for the realization of WMNs. Some of the differences between WMNs and ad hoc networks are outlined below. (a) The mesh routers in WMN form the backbone which provides large coverage, connectivity and robustness. But in ad hoc networks, the connectivity depends on the individual contribution of end-users. (b) The gateway and bridging functionalities in mesh routers provide the integration of WMN’s with other networks such as Internet, cellular, IEEE 802.11, IEEE 802.15, IEEE 802.16 and sensor networks. Unlike ad hoc networks, the routing and configuration functionalities of the mesh routers reduces the load on end-user devices. (c) The mesh routers can be equipped with multiple radios to perform routing and access functionalities which improves the capacity of the network. On the other hand, ad hoc networks use same channel for routing, network access, etc., which result in poor performance. (d) Unlike in WMNs, we run into several challenges with routing protocols, network configuration and deployment in ad hoc networks because its topology depends on the movement of users.
The mesh network is dynamically self-organizing and self configuring, with the nodes in the network automatically establishing and maintaining connectivity among themselves. These features provide many advantages for WMN’s like good reliability, market coverage, scalability and low upfront cost. WMN gained significant attention because of the numerous applications it supports, e.g., broadband home networking, community and neighborhood networks, delivering video, building automation, in entertainment and sporting venues, etc. Currently, hotspot IEEE 802.11 WLAN deployments are prevalent across coffee shops. A couple of obvious problems with this deployment is the location of access points and the presence of dead zones without service coverage. Though site surveys can be done to eliminate dead zones, it is very expensive. Installation of multiple access points can also be prohibitive cost-wise. The issue with accesspoints can be resolved by replacing accesspoints with mesh routers. In WMN, the mesh routers cooperatively route each others packet to destination. This results in flexible communication. The issue with dead zones can be eliminated inexpensively by adding more mesh routers or by changing the power level or location of mesh routers. The other wireless networks are not capable of multi-hop networking and hence mesh network is most suited for coffee shops, airports, hotels, etc.
In WMN, the mesh clients can access the network through mesh routers or directly via other mesh clients. To support end-to-end communication, effective routing algorithms are required to find high-throughput paths between source and destination. However, it is more difficult to find paths in wireless networks, as compared to wired networks. This is attributed to many factors. First, the channel errors in wireless links make them unreliable. Second, the communication links break when nodes move out of their transmission range. Third, achievable channel rates may be different in different links because of the dependence of the link quality on path loss and distance between the neighbors. Finally, the interference in the wireless medium from other ongoing simultaneous wireless transmissions plays a significant role.
In general, there are many routes between each pair of nodes in a network. Each route use different set of links with different throughput. The routing protocol should select a path with high-throughput. Routing protocols use route metrics to decide the best route between a pair of nodes [4]. To perform efficient routing, good routing metrics are required for path computation. Hop count is a widely used metric in both wired and wireless networks. For wired networks, shortest path is assumed to the path with minimum delay, and therefore hop count is a good cost metric. However for wireless networks, minimum hop count is not an accurate performance metric because it could cause congestion problems and power depletion at some specific nodes. Since WMN’s share common characteristics with the ad hoc networks, the routing protocols designed for ad hoc networks can be applied to WMN’s. On the other hand, the static nature of mesh routers and the mobile nature of client nodes implies that the routing protocols for ad hoc networks may not be suitable for WMNs. Based on the specific requirements of WMN, we believe that a good routing protocol should find paths with minimum delay, maximum data rate and low levels of interference. Also, an effective routing metric must be able to capture the following characteristics of the links accurately.
- •
Packet loss ratio. An effective routing metric should capture the packet loss ratio because lossy links can result in transmitting a packet multiple times on a link which inturn degrades the throughput and maximize the end-to-end delay of the flow in the link.
- •
Link capacity. In wireless network, the maximum transmission rate between a pair of nodes depends on the distance between the pair of nodes unlike in a wired network. Hence when the distance between two nodes increases, the channel quality degrades which inturn results in low link capacity. Therefore a good routing metric should favor a path with higher link capacity.
- •
Interference. Interference among wireless links have a serious impact on the performance of multi-hop wireless networks. Hence a good routing metric should take into account the interference of the wireless links. Two types of interference exists in wireless networks, intra-flow and inter-flow interference. Intra-flow interference is the interference caused when the nodes on the path of same flow contend with each other for the channel bandwidth whereas inter-flow interference is the interference caused when the nodes on the adjacent path contend with each other for channel bandwidth. Using multiple channels in multi-radio WMN can greatly improve the throughput of the network [5], [6]. Fig. 2 shows the performance of multiple channels in multi-radio WMN as opposed to single channel wireless network.
In this article, we propose a new routing metric called as Interference-Load Aware metric, ILA. This metric can be used to find paths between the mesh routers. The mesh clients do not need to participate in a routing algorithm since clients always send data to their respective routers. Only the routers are involved in the path selection and determination. The aim of the proposed metric is to find paths with less congestion, minimum packet loss, low level of interference and high data rate. Towards the end, the mesh routers are required to keep track of the traffic load on themselves, as well as their neighbors. The traffic load of the neighbors is a potential source of interference and paths with high interference should be avoided. The simulation results show that ILA provides better performance than the existing routing metrics for Wireless Mesh Networks.
The rest of the article is organized as follows. Section 2 presents an overview of the existing routing metrics for ad hoc and wireless mesh networks. Section 3 presents the design of the proposed interference aware routing metric, ILA. In Section 4, we present an overview of the Ad Hoc On-Demand Distance Vector Routing (AODV) protocol and discuss several implementation issues like Expected Transmission Time (ETT) measurement, load estimation of the interfering neighbors, etc. Section 5 describes the simulation setup and the performance results. Section 6 concludes our work and outlines our future directions.
Section snippets
Related work
A good routing metric should find paths with links that have high data rate, low loss ratio and low level of interference. In this section, we describe the need for a new interference, load aware routing metric for multi-channel Wireless Mesh Networks by presenting an overview of the various routing metrics such as hop count [7], RTT [8], ETX [9], ETT [10], WCETT [11], MIC [12], [13], iAWARE [14] proposed for multi-hop wireless mesh networks. This section discusses the details and limitations
Design of Interference-Load Aware routing metric (ILA)
In this section, we present the proposed Interference-Load Aware routing metric, ILA. This metric addresses the aforementioned limitations of existing metrics such as hop count, ETX, ETT, WCETT, MIC, iAWARE for Wireless Mesh Networks. This routing metric finds paths with less congestion, low level of interference, low packet drop ratio and high data rate. MIC [12], [13] captures inter-flow interference by scaling up the ETT of the link by the number of interfering neighbors. The IRU metric of
Implementation details
In this section, we describe the details about operation of AODV protocol and several implementation issues for the ILA like ETT measurement, load estimation of interfering neighbors, etc. The proposed metric was incorporated in the AODV implementation in ns2 [19].
Performance evaluation
The performance of the proposed ILA is compared with ETT, WCETT and MIC using ns2. The performance is evaluated in terms of network throughput, average end-to-end delay, packet loss rate, sensitivity of metric to varying interfering traffic and routing overhead. In the case of Wireless Mesh Networks energy constraint is not an issue [1] and hence we have not discussed it in this article.
Conclusions
In this article, we present a new routing metric for multi-hop wireless mesh networks. This metric is based on the load on interfering neighbors and link transmission rates. We integrated this metric in the well known AODV routing protocol and compared to existing routing metrics for Wireless Mesh Networks. We presented a simulation study that showed how this metric outperformed the existing routing metrics. Our future work is to investigate the performance of these existing routing metrics in
References (21)
- et al.
A survey on wireless mesh networks
IEEE Communication Magazine
(2005) - et al.
The nominal capacity of wireless mesh networks
IEEE Wireless Communications
(2003) - et al.
Coverage and capacity of a wireless mesh network
Wireless Networks, Communications and Mobile Computing
(2005) Computer Networks
(1996)- A. Raniwala, T. Chiueh, Architecture and algorithms for an IEEE 802.11 based multi-channel wireless mesh network, in:...
- P. Kyasanur, N.Vaidya, Multi channel wireless networks: capacity and protocols, in: Tech. Rep., University of Illinois...
- et al.
Dynamic source routing in ad hoc wireless networks
- A. Adya, P. Bahl, J. Padhye, A. Wolman, L. Zhou, A multi-radio unification protocol for IEEE 802.11 wireless networks,...
- et al.
A high-throughput path metric for multi-hop wireless routing
ACM Mobicom
(2003) - et al.
Routing in multi-radio, multi-hop wireless mesh networks
ACM Mobicom
(2004)