Advertisement
Product Releases
Advertisement

Reducing Congestion on Wireless WANs

Mon, 12/02/2002 - 5:38am

By Chris Bogdon and Mark Ferguson, Padcom Inc.

The wireless marketplace in the United States is in the midst of unprecedented growth. While the introduction of new technologies will further contribute to the expansion of this market, the challenge will be to ensure that networks are used efficiently. If this does not happen, networks will become oversaturated with packets, resulting in decreased application response time and increased packet loss. For wireline networks, Ethernet switches are usually a viable means of keeping network congestion in check. Congestion on a wireless network, however, is more complex. To maximize the usefulness of the wireless network and keep customers satisfied, congestion must be kept to a minimum.

Since many wireless networks are migrating from proprietary interfaces to IP standards, users can now employ wireline applications in a wireless environment. Although this is great for application developers and end users, it can have a damaging effect on the network. Many applications, like web browsers, are the standard interface for applications. While the applications work well with 100MB Ethernet networks, wireless infrastructures are not suited to handle the increasing amounts of web content. This leads to network bottlenecks that slow user productivity and increase user frustration.

Wireless Network

Speed Range

Satellite

2,400 Bps - 9,600 Bps

Mobitex  (Blackberry)

8,000 Bps

CDPD

9,600 Bps - 19,200 Bps

Private RF Technologies

4,800 Bps - 19,200 Bps

Motorola RD-LAP

9,600 Bps - 19,200 Bps

DataRadio

9,600 Bps - 32,000 Bps

M/A-Com OpenSky

19,200 Bps

GPRS

56KBps

CDMA 1xRTT

144KBps

Wireless LAN

1 MBps - 54 MBps

As the above chart shows, the majority of wireless networks are still less than 19,200 Bps. Although 3G wireless networks are becoming faster, they are still slower than most wireline networks. One means of easing the complexity of wireless networks and reducing the possibility of congestion is to combine a lower-speed cellular network with a higher speed alternative, like wireless LAN, creating a single virtual network. When mobile users are within range of a specified wireless LAN, applications will be offloaded to this network, thereby reducing congestion on the cellular network.

While middleware companies have been creating intelligent multiple network solutions since the mid 1990s, carriers and other service providers are just now starting to realize the need for this type of solution. Cellular carriers are embracing wireless LANs to reduce network congestion and wireless data users are taking advantage of the benefits of using multiple technologies. With the growth of wireless communication, solutions of this nature can reduce network congestion.

An important factor when developing a multiple network solution is the strength of the out-of-range detection algorithm. Switching between two types of networks requires an action that predicates the physical switch in networks to occur. In most cases, the switch will be initiated any time the user goes out of range of one wireless network. However, how does the roaming algorithm detect the moment the transceiver has left the coverage? Also, how can a multiple network roaming solution use other types of criteria than just plain network coverage?

One approach to solving the problem of wireless roaming is to apply the solutions used on wireline networks. One such solution is to run server applications that can determine if a client's computer is actively communicating with the server. The server (or client) will send end-to-end keep-alive packets or broadcasts to detect when the user is no longer communicating. If the user doesn't respond to this keep-alive message, it can be assumed that user is no longer using the network. The downside of this solution, however, is that it creates unnecessary packets on the network: the more keep-alive messages being sent, the more congested the network becomes.

As mentioned earlier, Ethernet switches are another way to reduce congestion on wireline networks. The switches are placed between client and server computers to provide dedicated bandwidth. Unfortunately, a solution like this cannot work on a wireless network, since all users are, in effect, sharing the frequencies used during communication sessions. There is no method of allocating a single channel to give a user a virtual bandwidth pipe without affecting the other users on that frequency.


For wireless networks, a better solution is to address the problem of the roaming detection algorithm. One way to do this is to intelligently query the transceiver to determine if a user should roam into a different network. Normally, this process requires a close coupling between the application and the transceiver used to access the network. One of the benefits of querying the modem is that the application can determine critical status information without incurring network overhead. In the minimal configuration, an application would be able to determine the received signal strength of the current base station. Collecting information this way does not produce any network overhead. Therefore, the application can determine instantly whether the user is in range. In other advanced uses, signal strength can be used to provide input to advanced algorithms to approximate when a user will lose connectivity. Applications can be written to statistically determine when that user will be out of range based upon his current behavior.

The ability of some modems to determine the network registration status -- in other words, if a user is in a home or roaming state - further enhances the ability of this solution. These parameters create improved roaming algorithms, resulting in a better experience for the end user. A case in point is setting the parameters to allow mobile users to automatically switch to a less expensive network when possible.

Due to the unique characteristics of wireless networks, specialized solutions like modem querying must be developed to address the problem of network congestion. These solutions will ensure the scalability of wireless networks as the industry continues to grow.

Chris Bogdon is Chief Technical Strategist at Padcom, Inc. He is responsible for helping determine the future direction of Padcom products, and helps to identify synergies between Padcom's products and other companies in the industry. He has been with Padcom for over seven years and has a B.S. in Computer Science from the University of Pittsburgh. He can be reached at cjbogdon@padcomUSA.com.

Mark Ferguson is Director of Marketing & Strategic Planning at Padcom, Inc. and is responsible for forging business relationships and maintaining the company's media and advertising presence. He has been with Padcom for two years and has a M.A. from Lehigh University. He can be reached at mferguson@padcomUSA.com.

Advertisement

Share this Story

X
You may login with either your assigned username or your e-mail address.
The password field is case sensitive.
Loading