期刊名称：International Journal of Hybrid Information Technology
印刷版ISSN：1738-9968
出版年度：2008
卷号：1
期号：2
出版社：SERSC
摘要：Neighbor attachment is the process of establishing links between nodes that on the physical layer are visible to each other. In ad-hoc and sensor networks many strategies can be followed to connect to adjacent nodes. A simple scenario would be to consider all nodes within the transmission range of a node as its neighbors. More sophisticated scenarios can establish neighborhood relations with a limited number of carefully selected nodes. Different neighbor attachment policies result into different network topologies at the link-layer. In any topology a packet sent from a node to one of its neighbors will experience interference, which is caused by simultaneous communications between other nodes in the network. Obviously, high levels of interference would severely disrupt communications and reduce overall network capacity. In this paper we examine how neighbor attachment policies can change the amount of interference experienced by network nodes. We will also show that Carrier-to-Interference ratio (C/I) is directly affected by neighbor attachment policies. Our study enables us to identify neighbor attachment directives that can reduce interference, or optimize C/I in large multi-hop ad-hoc and sensor networks. 1. Introduction Neighbor attachment in ad hoc and sensor networks directly affects the link layer networktopology. By adjusting some controllable parameters, such as transmission power and antennadirection, different topologies can be formed. A properly designed topology can reduce powerconsumption, remove low quality links, and improve routing efficiency in a multi hopnetwork. Especially innovative physical layer techniques, like antenna beam steering andbeam forming, introduce new possibilities for neighbor attachment. For example, by usinghigh gain directional antennas, more neighbors can be detected in a specific direction, whileat the same time interference with nodes outside the main antenna lobes are reduced.The state of the art literature related to topology control focuses mainly on life timeextension, hop count reduction, or performance optimization (e.g. [1]), without taking intoaccount effects of neighbor attachment on the amount of interference experienced in thenetwork. In this paper we attempt to fill this knowledge gap by comparing four widely usedneighbor attachment policies based on their interference imprint. The motivation for our workis the fact that interference has a direct impact on the capacity of a network. Consequently,our study paves the way to identify the suitability of each neighbor attachment policy tosupport different application types in ad hoc and sensor networks.The structure of this paper is as follows. In Section 2 we provide an overview of relatedwork in this research domain. In Section 3 we describe the neighbor attachment policies var currentpos,timer; function initialize() { timer=setInterval("scrollwindow()",10);} function sc(){clearInterval(timer); }function scrollwindow() { currentpos=document.body.scrollTop; window.scroll(0,++currentpos); if (currentpos != document.body.scrollTop) sc();} document.onmousedown=scdocument.ondblclick=initializeInternational Journal of Hybrid Information Technology Vol. 1, No. 2, April, 2008 34considered in this paper and show how they affect the network topology. Our findings relatedto the amount of interference and resulted Carrier to Interference (C/I) ratios for differentnetwork topologies are presented and discussed in Section 4. In Section 5, we summarize afew major conclusions.2. Related work Several papers have already studied interference in ad hoc and sensor networks. Here weprovide a helicopter view of main approaches in these papers, and position our own researchin relation to them. In [2] authors have looked at the impact of interference on theconnectivity properties of a dense network with random deployment of nodes. This line ofwork is pushed forward in [3] where authors discuss the effect of interference on the linkquality and connectivity of large networks. Although much attention is given to theconnectivity problem in ad hoc networks, effects of interference on capacity are onlydiscussed by approximation. Also, in the calculation of interference in [3], the dependency of var currentpos,timer; function initialize() { timer=setInterval("scrollwindow()",10);} function sc(){clearInterval(timer); }function scrollwindow() { currentpos=document.body.scrollTop; window.scroll(0,++currentpos); if (currentpos != document.body.scrollTop) sc();} document.onmousedown=scdocument.ondblclick=initializeInternational Journal of Hybrid Information Technology Vol. 1, No. 2, April, 2008 34considered in this paper and show how they affect the network topology. Our findings relatedto the amount of interference and resulted Carrier to Interference (C/I) ratios for differentnetwork topologies are presented and discussed in Section 4. In Section 5, we summarize afew major conclusions.2. Related work Several papers have already studied interference in ad hoc and sensor networks. Here weprovide a helicopter view of main approaches in these papers, and position our own researchin relation to them. In [2] authors have looked at the impact of interference on theconnectivity properties of a dense network with random deployment of nodes. This line ofwork is pushed forward in [3] where authors discuss the effect of interference on the linkquality and connectivity of large networks. Although much attention is given to theconnectivity problem in ad hoc networks, effects of interference on capacity are onlydiscussed by approximation. Also, in the calculation of interference in [3], the dependency ofinterference on the location of nodes inside the network is not considered. In [7] authorsrepresent an approach to preserve network connectivity based on the principle of maintainingthe number of physical neighbors of every node equal to or slightly below a specific thresholdvalue. They show that interference remains bounded under this approach. Here we show thatinterference always remains bounded in ad hoc networks regardless of the node degree. Thisfinding matches with previously found results in [4]. In [5], authors give algorithms toconstruct a network topology for wireless ad hoc networks such that the link interference inthe resulted topology is either minimized or approximately minimized. Having the sameobjective of minimizing interference, many works (e.g. [6]) have suggested construction oftopologies that require low transmit power. However, we show here that minimizinginterference on the link level is not the sufficient condition to guarantee maximum Carrier toInterference ratio (C/I), and consequently maximum capacity.In summary, topology control policies published so far do not take into account in arealistic manner the amount of interference experienced in the network. Due to the impact ofinterference on channel capacity, without proper interference estimation, the capacitycalculation methods provided in the literature lack accuracy as well.3. Neighbor attachment policies The neighbor attachment policies considered in this paper are described briefly here.Policy 1: Nodes are equipped with omni directional antennas. Any node considers allother nodes within its maximum transmission range (its coverage area) as itsneighbors.Policy 2: Nodes are equipped with directional antennas [8] with beam width ¦Á. Each nodesweeps its surroundings sector by sector, and establishes neighboring relationswith all nodes that are discovered in this way. Considering additional gain of thedirectional antennas in comparison to omni directional antennas, this policyginterference on the location of nodes inside the network is not considered. In [7] authorsrepresent an approach to preserve network connectivity based on the principle of maintainingthe number of physical neighbors of every node equal to or slightly below a specific thresholdvalue. They show that interference remains bounded under this approach. Here we show thatinterference always remains bounded in ad hoc networks regardless of the node degree. Thisfinding matches with previously found results in [4]. In [5], authors give algorithms toconstruct a network topology for wireless ad hoc networks such that the link interference inthe resulted topology is either minimized or approximately minimized. Having the sameobjective of minimizing interference, many works (e.g. [6]) have suggested construction oftopologies that require low transmit power. However, we show here that minimizinginterference on the link level is not the sufficient condition to guarantee maximum Carrier toInterference ratio (C/I), and consequently maximum capacity.In summary, topology control policies published so far do not take into account in arealistic manner the amount of interference experienced in the network. Due to the impact ofinterference on channel capacity, without proper interference estimation, the capacitycalculation methods provided in the literature lack accuracy as well.3. Neighbor attachment policies The neighbor attachment policies considered in this paper are described briefly here.Policy 1: Nodes are equipped with omni directional antennas. Any node considers allother nodes within its maximum transmission range (its coverage area) as itsneighbors.Policy 2: Nodes are equipped with directional antennas [8] with beam width ¦Á. Each nodesweeps its surroundings sector by sector, and establishes neighboring relationswith all nodes that are discovered in this way. Considering additional gain of thedirectional antennas in comparison to omni directional antennas, this policygenerally results into more discovered neighbors (at farther distances) than thefirst policy.
