Traditional wireless services such as cell phones, radio and Wi-Fi are designed using a hub-and-spoke model where the hubs are deployed and interconnected as part of a well planned structure. In contrast, urban grids based on mesh networking emerge opportunistically from existing wireless hardware, without much prior planning. Packets skip from device to device, seeking one that is connected to a wireless access point from which to enter the Internet. Thus, in the urban grid the user devices become the network. This radically new model of topology growth/evolution and of network operation poses unique challenges at every level. In this project we plan to address these challenges.

In order to better focus our study, we first consider some representative applications of the urban mesh network. From these applications we then derive the research themes and goals that will drive this study.

Figure 1: Urban Grid and Intelligent Transport System.

In summary, all the above applications and operational modes require advanced radios and novel network protocols. Current protocols for the most part enable Internet access (e.g. music download, path finding, etc.). The critical link is the wireless link from the station to an Internet resource and the main challenge is the selection of the best alternative for the last hop (e.g. IEEE 802.11, 3G, satellite, etc.). This problem, also known as the vertical handoff problem, has been already studied extensively in the literature [StK98].
However, an increasing number of emerging scenarios involve mesh networks and a multihop ad hoc wireless network is much more cost effective than traditional cellular type systems if the nodes are in proximity to each other. For example, for car safety and more precisely accident prevention it behooves a car to broadcast in the ad hoc network its position/velocity/acceleration (and possibly other parameters and environment information) so that two cars on a collision course can take evasive action. Passengers in cars going in the same direction may engage in distributed video games, chats and impromptu videoconferences to review notes before the meeting, etc. To this mobile mesh network pedestrian cellular phone users may also connect using IEEE 802.11, for example. More generally, one will notice that the last hop and multihop model must coexist in most of these applications. For example, if a boulder has obstructed the road, this information must be propagated into the mesh network as well as via 3G or via long range radio links, so that the information quickly reaches nodes many hops away. Mesh networks have a number of other benefits. First of all, they are highly adaptable and scalable. Mesh networks can have redundancy built-in so "each device has two or more paths for sending data". This redundancy makes mesh networks very reliable. For example, in case of a collapse of the fixed infrastructure, the back up would be the mesh network supported by the thousands of cars on the road (driven or even parked).

In this project the focus will be on the opportunistic coexistence and balance between mesh networking and fixed infrastructure communications. We will design and test transparent, adaptive MAC, routing and transport protocols. The lessons learned here will be transferable to many other systems consisting of large numbers of units in motion. For instance, the Campus environment where students equipped with PDAs can establish opportunistic mesh networks to extend wireless LAN reach or to support group communications. Or, the automated battlefield where manned units are interconnected with unmanned UAVs. UCLA is engaged in both Campus and battlefield networking studies. Our study will leverage the results of those projects. The urban mesh environment however is unique for the richness and breadth of the applications and the challenge in designing the radios (eg, radios equipped with MIMO antennas) which must be powerful, flexible, reconfigurable, robust and yet economical. The breadth of applications and the flexibility and richness of the radio options translate in turn into a set of demanding requirements for the network protocols, from MAC to middleware.