Context discovery in ad-hoc networks
Fei Liu · 2011
With the rapid development of wireless technology and portable devices, mobile adhoc networks (MANETs) are more and more present in our daily life.Ad-hoc networks are often composed of mobile and battery-supplied devices, like laptops, mobile phones, and PDAs.With no requirement for infrastructure support, MANETs can be used as temporary networks, such as for conference and office environments, and for disaster areas.The disadvantage is that they usually have limited bandwidth and that devices in ad-hoc networks have energy-constrained power supplies, which requires simple and efficient underlying communication protocols.One of the most fundamental actions that such devices in networks need to do is to find information about the environment they are operating in.To share and use the available context information in the network, devices first need to discover and locate the required information.This action is called context discovery.However, none of the existing discovery protocols can well support resource-limited fully-distributed MANETs.Therefore, in this thesis, we design and develop a new context discovery protocol for MANETs, which is called Ahoy.By using attenuated Bloom filters, Ahoy reduces traffic load to discover available context information and provides directional probabilistic querying.We build an analytical model to evaluate the performance of Ahoy compared with two conventional approaches: pro-active and reactive discovery protocols, and to allow for optimization of Ahoy's parameters.The results of the analytical model are validated by simulations.We estimate the network traffic generated by Ahoy in both static and dynamic environments.We find that Ahoy requires significantly less network traffic than the other two protocols in static networks, and that it is stable in a dynamic environment in which the network topology may change.We also study the vulnerability of Ahoy when it encounters different malicious attacks.Our analyses shows that compared with pro-active and reactive protocols, Ahoy is not more vulnerable than the other two protocols.In some cases, the use of attenuated Bloom filters can even help to protect the contents of packets up to a certain level.In case of serious risks, we propose specialized security countermeasures to enhance the network security of Ahoy.Finally, we build a prototype of Ahoy and test it on UNIX-like platforms.v vi Through these analysis and studies, we conclude that the novel discovery protocol Ahoy proposed in this thesis can discover information efficiently, while generating only little network traffic, in both static and dynamic fully-distributed MANETs.This chapter is organized as follows: Section 1.1 introduces background information.The motivation for the thesis is presented in Section 1.2.Then, we discuss the design requirements and assumptions, in Section 1.3.Thereafter, we pose the main research questions in Section 1.4.Section 1.5 elaborates on the approach and the structure of the thesis. 1 2 1.1 Background Wireless technology is developing rapidly, and in recent years, it has been deployed in many different application areas, from personal devices to satellites.We can connect to almost every device using wireless technology.More and more consumers possess personal devices, such as PDAs, laptops, and cell phones, which are facilitated by wireless communication technology, such as Bluetooth [7] and WIFI [39].As a result, research and application developments are extending from the traditional wireless access networks to networks with a more direct communication manner: mobile ad-hoc networks (MANETs).MANETs, unlike Ethernet and infrastructure Wireless LAN (WLAN), do not rely on fixed infrastructures.Devices can establish an arbitrary network via wireless communication when needed.We call the devices that establish the network, the nodes of the network.The wireless communications between nodes are called links.Generally, nodes are not required to load or exchange configuration files to form or join the network [73].Often nodes are battery-powered and able to move freely in any direction at any speed, which can lead to a frequent variation in connectivity.Wireless technologies that are used in MANETs, like Bluetooth, have improved in recent years.However, they still cannot provide the same data rate and bandwidth as Ethernet and infrastructure WLAN.MANETs can be used in various situations, especially when no infrastructure support is available or when it is too time-consuming and expensive to set up an infra-structured network.Normally, MANETs are temporary networks and their topology is unpredictable and dynamic.Typical scenarios in which MANETs are established include [71]:• Office and conference centers.In such an environment, most of the resident devices, such as desktop computers, printers, and scanners, are generally connected to Ethernet or WLAN.However, the mobile devices from employees or visitors, are often not authorized to connect to the fixed network.Consider a meeting scenario, where many visitors coming from cooperating partners need to share documents, exchange name cards, and use printers.Ad-hoc networks can satisfy such needs.• Disaster relief areas.One of the major applications for ad-hoc networks How can nodes in a context-aware ad-hoc network find and locate requested context information types fast and precisely, with limited bandwidth usage and power consumption?We should cope with the following research topics to resolve the main research question during the design.Research Question 1: Protocol design.How to discover context information to fulfill the mentioned design requirements?RQ 1.1: Context representation.-What is the proper manner to represent context information during the discovery process?-How should we record the information availability in the network?Which nodes, if any, may keep lists or directories of available information.What is the best choice for our situation?RQ 1.2: Discovery method.How to find information in a fully distributed network?We want to announce and query the information in a manner that does not generate a large amount of traffic.In general, a fast discovery protocol requires the announcement of detailed context information, which consumes a significant amount of bandwidth and battery power.How can we obtain an efficient protocol, which at the same time limits the consumption of bandwidth and battery power?Research Question 2: Protocol performance.What is the performance of the new protocol, especially, in terms of the following aspects?RQ 2.1: Complexity.What is the complexity of our protocol?It is important that the protocol itself is not too complex.Complex algorithms may consume too much power for computation and transmission.RQ 2.2: Scalability.How does the protocol perform under different network scale, i.e., small, middle, or large networks, with different network densities?Does the protocol have a relatively reasonable performance in high density networks?