[thought] Mesh Networks

Problem Area 2: Mesh Networks

 

    Mesh networks are a relatively new in commercial networking though they have been used in the military for quite some time. The entire concept of such networks is based on mutual co-operation between subscribers. An example might help to motivate the concept.

 

     Let us assume Bob has recently subscribed to a cable modem service that can provide bandwidths (both uplink and downlink) in excess of 100 Mbps (a very realistic assumption in the future). For this service Bob pays the ISP a fixed amount of $100/month (or less). Bob has a wireless-access-point-plus-cable-modem that serves as base station for him throughout his house. Adam and Chris are very close friends and also next-door neighbors of Bob. When they see Bob?s broadband service, they naturally want to get one too, but Bob realizes that since he does not use all of 100 Mbps at any given time, he might as well share his connection with his neighbors, who can also share the cost for his service.

Immediately the three decide that this is a great idea, and buy a cheap mesh router(MR) easily available in the market and install it in each of their houses. These MR?s once set up automatically communicate with each other and configure themselves to form a network that can cover all the three houses. Now all three of them can get very good connectivity and all are happy. Moreover, the MR comes with software that can identify the traffic distribution from all the three houses and give the three friends an option of dividing the monthly cost.

 

     The above is a simple but valid example of mesh networks, and it can be extended to cover entire building communities that provide wireless access to its tenants at the fraction of the cost. A more complex application might be start-up ISP?s providing seed nodes in communities, and people can join in by installing an MR on their rooftops. Thus by mutual co-operation everyone benefits. Also such networks are highly adaptive and do not depend on the vagaries of one rogue subscriber. In other words, even if one or two MR?s are switched off, the MR?s will reconfigure themselves to form a new configuration and still maintain connectivity.

 

    Till date, hardly 3 or 4 introductory papers have been published in IEEE on wireless broadband mesh networks, though many start-up companies have mushroomed claiming their products can efficiently support mesh networks. Thus lack of documentation/standards seems to be a major hurdle for research into this area.

 

Some of the research issues are:

• Standardized protocols for auto-configuration of MR?s: Some standardized protocols need to be established that can help MR?s to communicate with each other and set up point-to-point links with their neighboring MR?s. Also routing (or switching or relaying) functionality must be aptly supported by the MR?s, such that all the paths can efficiently lead to seed nodes (or nodes that have a wired interface to the Internet).
• Supporting Mobility: Supporting mobility in such a mesh network, will be a major issue. In other words, intra-MR mobility should be complemented by inter-MR mobility. Newer network architectures will need to be established to achieve this goal.
• QoS issues: It will be necessary to provide QoS guarantees to real-time traffic. The network design should include considerations for QoS issues right from the planning stage. QoS support must be deeply embeeded in the network design rather than forcefully overlaid on the final product.
• Self-Organizing and Self-healing: Algorithms need to be researched for providing both these properties. The self-organizing property allows the network to be configured in an ad-hoc manner such that both new and old nodes are automatically configured with zero user intervention, while the self-healing property, allows the network to respond to different types of failures