Wireless networks offer several advantages such as easy and fast deployment over conventional wired networks. To provide large cover range wireless networks such as MAN, there are three kinds of technologies. The first one is enabling the delivery of last mile wireless broadband access like IEEE 802.16. The second one is deploying a high density of wireless local area networks such as IEEE 802.11 [1]. The last one is a wireless mesh network which intends to support a broad range of deployment scenarios and data delivery over self-configuring multi-hop topologies.
IEEE 802.11s [2], an extension for mesh networks of IEEE 802.11, specifies a framework using the IEEE 802.11 MAC/PHY layers to support broadcast, multicast and unicast data delivery by data forwarding and path selection.
In the infrastructure basic service set (infrastructure BSS) of IEEE 802.11, a station (STA) accesses Internet through an access point (AP) which forwards the data to the destination. In each infrastructure BSS, only a single channel is needed for the STAs to communicate with the AP. However, in IEEE 802.11s, there are two kinds of traffic, BSS traffic which is forwarded by the AP to the STAs and mesh forwarding traffic which is forwarded by the intermediate mesh points (MPs) on the path. Thus, a mesh access point (MAP) must deal with these two kinds of traffic simultaneously. In a single-channel scenario, BSS traffic occupying the channel potentially starves the neighbor MPs and results in long packet delay or serious packet loss. Therefore, separating different traffic into different channels is an intuitive thought. In a multi-channel scenario, it can not only separate BSS traffic and mesh forwarding traffic into two channels but also separate mesh forwarding traffic into multiple channels to improve network throughput.
In the IEEE 802.11 wireless networks, each channel applies for one radio.
Representing that, channels need radios on each wireless device for the best performance. Nevertheless, the more radios, the more hardware cost and power consumption, especially for mobile stations. Moreover, a lot of devices were designed for a fixed number of radios and they are not easy to modify to equip with more radios. When radios are less than channels, switching among channels results in packet loss and link disassociation in unattended channels. Such a device whose radios switch among channels is called a switch node like a MAP or a MP in the multi-channel mesh networks. To take a MAP for an example, link disassociation makes its neighbor MPs rediscover the routing path and forces the STAs associate with another AP. While the radio switches back to the channel, the switch node rejoins the topology of the mesh network and the STAs might re-associate with it because of a better link quality. The above may exhaust the network resources if it happens frequently.
N N
Several researchers have studied the multi-channel protocols. They discussed that each node requires one radio [3] [4], or two radios [5] [6] [7], or radios [8] [9]
for a N-channel environment. All the methods require that each node supports their specific protocols which might not be compatible with the standard. In addition, using a single radio in a multi-channel mesh network requires one unrealistic constraint, the global timer synchronization, which is very difficult to achieve in the mesh networks because the delay in a large area cannot be correctly estimated. Another work dealing with channel switching [10] provides a method for a STA to connect to multiple networks with a single radio, but it is insufficient for the mesh networks because there are more scenarios like channel switching for a MAP and equipping with multiple radios.
N
Although much research has been devoted to the multi-channel wireless
networks, little information is available on the IEEE 802.11s mesh networks. This paper investigates how to use less radios to switch among channels in the IEEE 802.11s mesh networks. The proposed method, Time Ratio Aware Switching Superintendent (TRASS), could run on all devices in a mesh network and handle channel switching. This method includes two parts: (1) TRASS mechanisms existing in IEEE 802.11 and IEEE 802.11s ensures frame delivery to the switch node. For the infrastructure BSS, a switch node uses the Hybrid coordination function Controlled Channel Access mechanism (HCCA) [11] or the CTS-to-self mechanism [12] to notify the STAs to wait for it. For the mesh networks, a switch node enters its power-saving mode [2] to notify its neighbor MPs to buffer the data; (2) a TRASS algorithm considering the general case that is a switch node has only radios in a
M-channel environment while
N
M is greater than N. This algorithm selects which
channel to stay in according to the traffic load ratio for the switch node and adaptively allocates time for the channel by its total traffic load ratio. This paper carries out simulation-based and implementation-based studies. It examines packet loss ratio, channel utilization, throughput and average latency. Additionally, the relationship between these factors and the number of radios in the multi-channel mesh networks is also investigated.The rest of this paper is organized as follows. Chapter 2 introduces the IEEE 802.11 MAC, the architecture of the IEEE 802.11s mesh networks and the impact of channel switching with less radios in a multi-channel environment. Chapter 3 presents TRASS and illustrates its detailed operations with some examples. Chapter 4 shows the system architecture of implementation and chapter 5 investigates the evaluation results of simulation and implementation. Finally, we conclude this work with some future direction in chapter 6.