Your browser is not supported. Please update it.

31 October 2005

Microsoft digs more deeply into spontaneous mesh

Microsoft Research has released a new software tool which allows a PC to connect to multiple Wi-Fi networks simultaneously, the latest step in its bid to use ad hoc wireless mesh to drive mass market uptake of Windows devices and act as a Trojan horse to take its systems into new markets.

With VirtualWiFi, a single wireless card can appear as multiple virtual WLan cards, which can be configured by the user to link to different access points or other devices.

The tool is only a prototype at this point but shows how Microsoft, despite withdrawing from the Wi-Fi devices market, remains determined to be in the forefront of stimulating usage of wireless Windows devices by enhancing the functionality of WLans when running with its operating systems. Other key projects in this direction include its work on ad hoc mesh networking.

The virtual WLan tool may be used in two other applications. The first, called Client Conduit, is for fault diagnosis and recovery in wireless networks. It provides a thin pipe of communication between disconnected clients and back end servers, which perform wireless diagnosis and recovery. The thin pipe is achieved by running VirtualWiFi on the still functioning clients. These spontaneously connect to the disconnected devices over an ad hoc network to send messages from them to the recovery servers ‘ another example of Microsoft’s intense interest in self-managing wireless networks, which stimulate usage because only very basic infrastructure is required, and no operator.

The second application of VirtualWiFi, called Slotted Seeded Channel Hopping (SSCH), increases the capacity of wireless ad hoc networks using orthogonal channels, and is at the heart of the Microsoft mesh activities, which it sees as an important way to bring broadband to underserved communities cheaply, drawing Windows devices in its wake. The application uses VirtualWiFi to virtualize a wireless card with as many instances as the number of orthogonal channels. It then connects each virtual wireless card on a different orthogonal channel for spontaneous connections, especially in mesh configurations.

While mesh is becoming an accepted way for WISPs to extend their coverage across metro areas, Microsoft takes a more radical approach than most of the current products available, focusing on peer-to-peer ad hoc mesh of a type scarcely used outside community projects. The software giant believes such meshes, backhauled by WiMAX or satellite, could bring broadband wireless access (on the back of Windows of course) at low cost to developing nations and very rural areas. The official objective is to help bridge the social and economic ‘digital divide’ ‘ the underlying agenda, of course, to make Windows, into which the new mesh capability can be built, ubiquitous, and help it fend off the attack of Linux in emerging markets.

In June, Microsoft Research revealed that it had partnered with seven US universities to develop its Mesh Connectivity Layer, a loadable Windows XP driver which is currently based on Wi-Fi but could also support WiMAX or other protocols that lend themselves to meshing. The company sees this as a method for remote communities to gain broadband access for low investment, allowing a single high speed connection to be shared among many users.

Victor Bahl, senior researcher at Microsoft’s networking research group, said a company or individual could decide to ‘invest $800 per month to lease a T1 line and connect the remote neighborhood for a fee’.

The Layer lets computers communicate via the mesh, ‘tricking’ them into thinking they are directly linked to the internet. The software creates a virtual network adapter that looks like a regular network connection to the computer, and uses a mechanism called Link Quality Source Routing, based on an IETF standard, to route data among computers in the neighbourhood, explained Microsoft. This approach conflicts with that of Intel by routing between the layer 2 (media access) and layer 3 (network) level. ‘We created a Layer 2.5 where our routing software sits,’ said Bahl, while Intel firmly believes in implementing mesh at Layer 2, to avoid difficulties in assigning IP addresses to devices that cannot be ‘seen’ at the network layer. This approach lies firmly under the auspices of the IEEE standards group, which addresses Layers 1 and 2, rather than the IETF, and already there is something of a political and cultural debate brewing in wireless over which of these communities should dominate the agenda.