MAGNeT Research Seminar Series 08 (MRSS08)

MRSS08-P1  

Introduction to Multi-hop Ad Hoc Networks and a Survey of Open Problems

Sanjay Srivastava

Synopsis

 

Definition of mobile ad hoc network

Key characteristics of a MANET are (a) wireless physical medium, (b) multi-hop data transmission, (c) mobility of the nodes, and (d) no specific infrastructure requirement.

 

Physical and MAC Layer Issues

Accurate physical layer modeling is required for estimating the performance of the network. In particular, multi-path fading, shadowing, and Doppler effects are important in estimating the signal propagation characteristics. Free space and 2-ray models are two basic models with applicability in 3-d open spaces and near flat earth respectively.

 

Efficient protocols (MAC protocols) are needed for efficient shared use of the wireless media among the nodes. A number of protocols have been designed and standardized, e.g. IEEE 802.11x, ZigBee etc. However, performance of these under multi-hop conditions is not very satisfactory. See the references for details of these protocols and their performance comparison.

 

Capacity of an ad-hoc network as a function of node density is an important measure. Theoretical results [refs] suggest that for static ad hoc networks, capacity goes as n-1/2.

However, once mobility is introduced, it has been shown that a constant throughput per node pairs can be achieved. This has important consequences on the scalability of MANETs.

 

Modeling of Ad Hoc Networks

Ad Hoc networks can be modeled as graphs with nodes represented as vertices and links represented as edges. A randomly deployed mobile network can be modeled as a random graph. These graphs display a number of interesting properties e.g. sharp phase transition from a disconnected to a connected graph as a function of link-up probability, and “small-world” properties.

 

Queueing network modeling of ad hoc network have also been successful in estimating the performance measures of these networks. We will have more detailed presentations on these later in this seminar series.

 

Interesting Problems

Many of the successful realizations of ad hoc networks are in the domain of vehicular networks and sensor networks. Architecture of these systems tend to be application specific. However, there are some problems which are generic in nature.

  • Routing: Current focus is on energy efficient and fair routing protocols. One also looks at QoS enabled routing.

  • Cooperation: Ad hoc networks are based on the idea of cooperating nodes. There are situations however, where nodes may not wish to cooperate in order to maximize their lifetime. There are interesting protocol proposals for implementing forced cooperation.

  • Localization: Nodes in a randomly deployed mobile networks need  to know their spatial coordinates in many applications. Challenges are in the area of accurate localization without the need of specialized h/w support.

  • Delay Tolerant Network: Networks which do not have full connectivity at any given time are very challenging to analyze. Even basic routing functions require a relook. A number of problems in such DTN are being actively studied.

 

Simulation

Experimentation on actual deployed systems are very difficult and expensive. Simulations, if done correctly, offer good insight into the protocol performance. Challenges are in the area of creating and validating “node mobility model”, implementing correct physical layer model and MAC protocols, and finally doing correct statistical analysis of the simulation output.

 

Protocol Stack Implementation

Good quality protocol stacks for ad hoc networks do not yet exist. There is a need for open-platform stack development. This will also help allow fill the gap between simulation and experimentation.

References

 

A number of primary references on the basic topics mentioned above are available on my Topics in Wireless Ad Hoc Networks course website. Click here to access the link.