How 4GW tackles today’s wireless challenges

wireless

Fourth Generation Wireless (4GW) is designed to overcome the constraints of older wireless deployments.

A decade and more ago wireless implementations were, of necessity, based on non-standard proprietary technologies. With the emergence of the initial IEEE 802.11 family of standards, manufacturers focused on interoperability. Wireless equipment prices plummeted and the concept became far more viable – particularly for domestic wi-fi communications. Typically, there are one or two access points (APs) and just a few users.

The simplicity and ease of deployment led to this consumer-grade wireless turning up in the workplace, where it was tweaked and adapted to try to accommodate the more robust demands of business. Third-generation wireless evolved to include a centralised controller, able to configure and manage the APs and bringing security and policy management to wireless communication. The essentials remained however: a single cell architecture with the client device roaming between cells.

A major challenge in what has become “conventional” wireless topology is the limited number of wireless channels available, so that they have to be shared and re-used. The resulting contention leads to co-channel interference and consequent loss of throughput. Signal strength relates to the amount of throughput: the closer you are to the access point, the better the signal. Further away, you can connect, but data throughput will most likely be significantly reduced.

Architectural challenges

An effective deployment aims to deliver a decent average throughput throughout the target area, so it is common practice to add additional APs. In the usual 2.4GHz band, there are three channels available so the network architect has, in effect, only three colours to play with and some co-channel interference is inevitable, resulting in packet collisions and loss of throughput.

The more devices connecting to the network, the less overall bandwidth is available. The contention curve of wireless is similar to that of Ethernet: if they send data at the same time, the packets collide and both are lost. A clutch of devices connecting concurrently can results in the loss of 90% of actual throughput; the residual 10% of connectivity has to be shared among the client devices. Schools are a prime example: 30 users log on simultaneously in one location, all demanding immediate wireless connectivity, and the system is often found wanting.

A third issue facing wireless network designers is a process known as control association: the client devices themselves govern which AP they connect to – or associate with. The idea is that as the user device associates with an AP on entering the building, then should moves around associating the most appropriate AP available to it. Unfortunately, a device often finds and AP and sticks with it, ignoring better options as the user physically moves around the site.

A “sticky client” reduces its data transfer capability and can affect the AP itself adversely, particularly if the client is an older, slower type. Also, when a client decides to roam between APs, there is a handoff period of perhaps 100 milliseconds where no data is being passed. This is no problem for email, but it can be enough to drop a latency-sensitive application such as a voice call. .

If three clients arrive to take advantage of an AP, one might expect that each receives one-third of the available airtime. In practice, while they receive an equal amount of bandwidth, there is “airtime unfairness”: an older, slower client will grab a significant amount of the airtime so markedly reducing the overall throughput of the AP.

4GW answers

The topology of what has become known as 4GW depends on the vendor, but an interesting manifestation is the “channel blanket” approach, comprising the “virtual” cell: the world is seen as one big AP, in effect, rather than individual APs. Meru Networks and Exicom promote channel blanket architecture for wireless network coverage – albeit in their own specific way. Meru was voted Gartners’ “most visionary vendor” for the past two years in the wireless arena.

Dave Crowder is Meru’s Senior Systems Engineer, UK & Ireland: “4GW relies on packet-level coordination between APs to avoid data collisions. This means that you can choose to deploy entirely on a single channel and extend coverage easily. You can add APs without having to worry about channel pattern planning as the APs are instantly incorporated into the overall coverage.”

Along with optimised coverage and avoiding co-channel interference 4GW, says Crowder, is also far easier to administer and maintain: “It offers a marked uplift in usability. Low-level coordination at the AP avoids client contention and so eliminates the vast majority of packet collisions, while maintaining overall throughput at close to the theoretical maximum, no matter how many client devices attach.” With such a promising specification, the education and health sectors are natural target customers.
The client in a 4GW network is unaware of the individual AP hardware devices on the walls. The decision to roam is given, instead, to the network which examines not only signal strength, but also loading and makes intelligent decisions as to which AP a user device should connect to at any one time. As the user moves around, the network reassesses its decisions. This coordinated handoff, says Crowder, ensures there is no loss of throughput so that voice callers, for example, will be unaware of user movement between APs.

4GW tackles the issue of airtime fairness by maintaining a higher level of aggregate throughput for every AP. While 802.11n explicitly allows backwards compatibility, 4GW does not allow APs to waste time as it services older clients: “You can’t have slower clients using up large amounts airtime and eliminating the benefits of investing in 11n in the first place. 4GW can mitigate that skew towards those older clients by ensuring the faster clients are not disadvantaged and that the older clients still get the airtime they need.”

When appropriately configured, an 802.11n wireless network delivers at least six times the bandwidth and improved – though less predictable – coverage. Crowder muses on the possibilities that we now have the technology to replace the flood-wired office with ubiquitous wireless: “There are not enough people with sufficient confidence in wireless so far, but the advent of 11n means we are getting much closer, delivering the equivalent of 100 Mbps to the desktop, reliably and security.”
While 11n still involves three channels in the 2.4GHz band, they can be twice as wide as before, obviating the need for complex channel planning. The traditional approach is to place an AP in a gap and then replan around it, whereas 4GW involves “painting” sufficient coverage without having to channel plan.

Crowder adds: “Word is spreading, with thousands of 4GW deployments [by Meru Networks and others] worldwide across a wide variety of vertical sectors, particularly education and healthcare.”

This post was written by:

Adam Malik - who has written 29 posts on IP EXPO ONLINE.


Contact the author

Leave a Reply

Advert

IP EXPO Event News

IP Leaders Comment