Building Rural Mesh Network Prototypes Computer Science Essay

Published Date: 02 Nov 2017

Such networks are built using high-gain directional antennas that can establish long-distance wireless point-to-point links. Some nodes in the network (called gateway nodes) are directly connected to the wired internet, and the remaining nodes connect to the gateway(s) using one or more hops.

In this paper, we present a novel channel assignment framework for rural WMN. Our channel allocation scheme allows concurrently transmitting or receiving data from multiple neighbours links, and continuously doing full-duplex data transfer on every link, with the aim of creating efficient rural WMN.

We consider any adjacent links at a common node with different channels to ensure that concurrent transmissions or receptions. We also consider any link in the network as made up of two directed edges in full-duplex data communication. To each directed edge at a node, we assign a non-interfering IEEE 802.11 channel so that the set of channels assigned to the outgoing edges is distinct from channels assigned to the incoming edges.

Channel assignment is shown to be NP-hard. We frame this channel allocation problem in terms of adjacent vertex distinguishing edge coloring (AVDEC). Detailed assignment results on grid topology are also presented and discussed

Introduction

Background and Motivation

Directional antennas and long-distance communication links WMN has started to be used in industrial applications for building a wireless backbone interconnecting remote plants .

Recently, the interest has been rising [22, 26] in the design of long distance mesh networks for rural areas, using IEEE 802.11 (WiFi) equipment.

Rural areas have populations with very low paying capabilities, especially in developing regions. Therefore, it is urgently required to minimize the infrastructure costs for these networks, which play an important role for realizing sustainable society .

As depicted in Fig. 1, a typical IEEE 802.11 based rural mesh networks consist of a cluster of villages connected with each other through point-to-point wireless links. Some special nodes in this mesh, called the gateway nodes, are connected to the wired internet. Other mesh nodes connect to the gateway nodes (and thus, to the rest of the internet) through one or more hops in the mesh.

(Rural mesh networks are characterized by a static topology and very long distance communication links between the nodes.)

Rural mesh networks have four major characteristics, 1) a static topology (a node in this network is a village), 2) each mesh router (node) is equipped with multiple radios, potentially enabling to transmit or receive from multiple neighbors links simultaneously and to do the full-duplex transmit in its wireless links. 3) use of high-gain directional and 4) long-distance point-to-point links (several kilometres) [17:design and evaluation]. These features distinguish these networks from most networks considered in prior literature on multi-hop 802.11 networks [].

A primary concern facing rural mesh networks is the capacity reduction of the backbone network caused by wireless interference

The presence of multiple radios per node makes it possible for the use of non-overlapping channels for the various links. However, there is still much interference between those transmission links without appropriate channel assignment due to limited non-overlapping channels. The IEEE 802.11b standard and IEEE 802.11a standard offer 3 and 12 non-overlapping channels respectively, and the reuse of the same channel in a neighborhood must be limited, as simultaneous transmissions over the same channel conflict, resulting in a decrease of the throughput.

A promising approach is to equip routers with directional antennas which could further improve the system throughput by means of reducing the interference between adjacent links thus allowing more synchronous transmissions in the network. Meanwhile, as the [Design] has demonstrated, a line-of-sight long distance 802.11 link with high throughput can be established via using high-gain directional antennas.

However, side lobes from the directional antennae create radio noise in the nearby antennas and may cause interference with other adjacent links at a node, therefore, concurrent transmissions or receptions [?] on the same channel cannot be operated.

Specifically, a node cannot simultaneously transmit and receive on its adjacent links on the same channel in certain protocol due to presence of near-field effect.

An obvious approach towards addressing this problem is to operate the interfering links in different non-interfering channels. In order to provide an efficient rural mesh wireless network, an appropriate channel allocation strategy needs to be designed to reduce the impact of interference and improve network throughput.

B. Related Work

In this section, we introduce some related work that motivates us to propose the channel assignment framework for rural mesh networks.

The Digital Gangetic Plains (DGP) project and MIT’s Roofnet [6] are typical operational examples of rural mesh network [7]. However these projects are mostly built using omni-directional antenna.

Our work on channel allocation is close to that of Raman et al. [16]. Raman et al. considered the channel allocation problem in rural mesh networks which proposed a MAC protocol called 2P. The protocol not only enhances throughput but also discussed the problem of interference between the multiple radios in each node.

The main limit of the proposed approach is that although each router is equipped with multiple radios, they are activated one at a time so as to avoid interference, thus limiting the advantages of the mesh architecture.

In addition, the 2P can only be operated on a bipartite graph. In a general topology that is not bipartite, 2P may not fully utilize the network capacity. This problem has been investigated in [?,?]. The main idea is to divide the networks into multiple disjoint bipartite subgraphs, each of which is assigned a different channel. Therefore, 2P can be properly scheduled within each bipartite subgraph without interference. The authors also proposed a heuristic algorithm to construct a bipartite topology of the network. In [novel scheduling for concurrent transmit], the authors proposed a novel scheduling algorithm that maximally create bipartite graphs for concurrently transmitting or receiving WMN.

In[improving routing in long dinstance], the paper discussed one important feature concerning routers for long-distance wireless mesh networks, which possess the full-duplex capability. A distributed embedded router is implemented to ensure that the receiver (transmitter) pairs operate on the same channel.

Several researches have been done on the channel assignment strategies in wireless mesh networks [5], [18], [14], [15]. However, these works are principally distinguished from ours in terms of network features, and thus the interference properties are analyzed differently.

C. Contributions

The above observations motivate us to design an efficient，and more importantly simple and easy-to-implement channel assignment scheme for rural WMN with directional antennas.

In this paper, we make several important contributions towards developing efficient algorithms to solve this problem. First, we describe the requirements to establish a point-to-point 802.11 link between two nodes of a given network graph (And interference model). We then formulate this optimization problem as the Adjacent Vertex Distinguishing Edge Coloring (AVDEC) [?] problem that achieves a channel allocation with the above properties on all links of the given graph. In this coloring framework, all data links that have the potential to be active simultaneously without interference. We present and discuss the detailed assignment results on grid topology. Our algorithms also work on any random topologies, and show lower…

Lastly, we propose a heuristic algorithm to find an approximate solution to the problem.

D. Organization

The remainder of this paper is organized as follows. The system model is described in Section II, and the problem formulation is presented in Section III. In section IV, we propose a solution framework for the AVDEC model channel allocation problem, and the theory demonstration is given in Section V. Finally, we conclude this article in Section VI.

2. System model

In this section, we describe the architecture of rural wireless mesh networks. We also describe how the nodes will communicate with each other and discuss the interference model.

Network Architecture

We consider a rural wireless mesh network with stationary mesh nodes, where each node (village) is equipped with multiple radios using directional antennas. Some nodes will be designated as gateway nodes that will connect to the wired Internet. The gateway nodes will connect to other mesh nodes through line-of-sight long-distance links to form the communication backbone of rural wireless mesh network.

As mentioned earlier, a link between any two nodes can be considered as made up of two directed edges in opposite directions. We assume that all radios operate in full-duplex mode, i.e. a node can be simultaneously transmitting and receiving on its adjacent links. Formally, the rural mesh networks can be modeled as an undirected graph, where each mesh node is equipped with radios with directional antennas. The edge in corresponds to a wireless link between nodes and , while we denote and to represent the two directed edges on the link . There are totally non-overlapping frequency channels in the system.

Directional Antenna and Interference Model

In this paper, we consider mesh nodes equipped with multiple radios and each radio uses a directional antenna with a fixed transmitting (receiving) direction. The directional antennas give a mesh router the ability to focus its transmission energy or electromagnetic beam on a given geographical region, and effectively lower the noise floor of nearby routers. More importantly, directional antennas increase spatial reuse significantly as routers are able to transmit simultaneously [?,?,?].

While the directional antennas are designed to transmit and receive in a specific direction the directionality of this radiation becomes effective only at longer distances from the sender. This is also called the near field effect. As a result, there is considerable leaked radiation (side lobes) within a short area from the antenna. Because of this effect, adjacent links at a node interfere with each other in certain communication modes. Specifically, at any node, simultaneous transmissions and receptions on the same channel are not possible since the transmissions will interfere with the receptions. This is called Mix-Rx-Tx interference [?,?].

Another observation (as pointed out in [11]) is that while Mix-Rx-Tx interference prevents simultaneous transmission and reception at a node, a node can synchronously transmit (or synchronously receive) on all its adjacent links without any Mix-Rx-Tx interference. This is called SynTx (or SynRx). In a recent work, Raman et al. [11] have proposed the 2P MAC protocol based on this observation. The protocol operates on graphs by switching each node between two phases: SynRx and SynTx. When a node switches from SynRx to SynTx, its neighbors switch from SynTx to SynRx, and vice versa.

However, for such simultaneous transmission or reception to be possible (on the same channel) on adjacent links, they need to a) set "appropriate" power of each of the links, b) avoid small angle separation between links, c) turn off MAC level immediate ACKs, and d) prevent carrier sense based back-off implemented in the hardware [10].

However, this may not be true in practice. There are some issues. First, apart from Mix-Rx-Tx interference, the point-to-point links can also overlap spatially because of side-lobe leakage of directional antennae. Reducing side-lobes requires paying a lot of attention to electro-mechanical detail, and such antennae are currently very expensive. Second, apart from the side-lobes, the antennae also have a near-field effect. Third, the authors know the exact locations of the nodes so the power levels of the links can be engineered with a careful link-budget analysis to reject the interfering transmission [design].) Thus, for more headroom in the link-budget analysis in [design], it is desirable if two links assigned distinct channels. Fourth, in the study of rural mesh networks, we have also assumed that nodes are connected only through point-to-point directional antennas. An alternate model, to reduce cost, would be that a node connects to several other nodes through one sectoral antenna. Sectoral antennas again introduce further spatial interference problems both between each other and with other directional links that overlap with them. Hence, a new distributed link scheduling protocol may be required that can allow spatially overlapping links to operate without interference.

In [interference and traffic aware], the authors devised two testing scenarios. In the first scenario, each interface on a multi-radio mesh router to use the same frequency band, while in the second scenario, both the interfaces are using channels 1 and 11 of the 2.4 GHz band, respectively. The results show using the same band on both radios produces an high level of interference, so called cross-talk, that coupled with the mechanics of the CSMA/CA channel access scheme exploited by 802.11 results in a significant performance drop. On the other hand using two different transmission band leads to a 50% performance increase.

In the realization of a rural mesh networks, it is therefore necessary to adapt the behavior of the protocols and implement devices that will take into account the particular features of this kind of network.

3. Problem Statement

One of the goals of the channel assignment problem is to ensure network connectivity. To fulfill this requirement, we adopt the rule that there is a link between two nodes if and only if they are within the transmission range of each other. In other words, the two nodes need to tune their radios to a common channel for data transmission and reception. In this way, each node can communicate with every one of its neighbors. After we give channels to these wireless links, each radio at a network node gets its own assignment. Thus the problem of channel assignment becomes assignment each link in the network a channel.

In this section, we formulate the problem of assigning channels to links in the rural mesh networks on which our proposed concurrently transmitting or receiving and full-duplex data transfer will operate.

An important observation in our design is that if two links are allocated independent channels, they can be scheduled independent of one another.

On the one hand, we assign non-interfering channels to adjacent links, in such a manner that allows adjacent links at any a common node can concurrently transmit or receive without interference, and on the other, we assign non-interfering channel to each directed edge, in such a manner that allows each node can do full-duplex data transfer. In other words, since all links incidence a common node use different channels and no outgoing and incoming edges share a common channel, there is no risk of side-lobe and Mix-Rx-Tx interference. Thus, a node can be simultaneously transmitting or receiving along all links, while a node can be simultaneously transmitting on a set of channels on its outgoing edges, and receiving on a set of channels on its incoming edges.

We now propose a framework that achieves a channel allocation with the above properties on all links of the rural mesh networks.

For ease of presentation we introduce the following definitions. Given an undirected graph , the corresponding bi-directional graph is the directed graph such that,, and for every undirected edge (), there are two directed edges and in . (Since each physical link shall be used for information transmission in both directions using different channels, we model the network links as bi-directional edges in the network graph.) We assume that represents the set of available colors. We denote a function : such that no two edges with the same colors are incident with the any node . Furthermore, for any edge , there are two directed and in in corresponding graph . We also denote a function such that the following constraint is satisfied: for any , the set of colors assigned to its incoming links, is distinct from the set of colors assigned to its outgoing links, . In other words, for any , . We would like to use the minimum value of for communication in the network under this constraint. Throughout this paper, we will use vertices and nodes interchangeably, as well as edges and links.