Wireles Networking is a practical guide to planning and building low-cost telecommunications infrastructure. See the editorial for more information....

Estimating capacity

Wireless links can provide significantly greater throughput to users than traditional Internet connections, such as VSAT, dialup, or DSL. Throughput is also referred to as channel capacity, or simply bandwidth (although this term is unrelated to radio bandwidth). It is important to understand that a wireless device's listed speed (the data rate) refers to the rate at which the radios can exchange symbols, not the usable throughput you will observe. As mentioned earlier, a single 802.11g link may use 54Mbps radios, but it will only provide up to 22Mbps of actual throughput. The rest is overhead that the radios need in order to coordinate their signals using the 802.11g protocol.

Note that throughput is a measurement of bits over time. 22Mbps means that in any given second, up to 22 megabits can be sent from one end of the link to the other. If users attempt to push more than 22 megabits through the link, it will take longer than one second. Since the data can't be sent immediately, it is put in a queue, and transmitted as quickly as possible. This backlog of data increases the time needed for the most recently queued bits to the traverse the link. The time that it takes for data to traverse a link is called latency, and high latency is commonly referred to as lag. Your link will eventually send all of the queued traffic, but your users will likely complain as the lag increases.

How much throughput will your users really need? It depends on how many users you have, and how they use the wireless link. Various Internet applications require different amounts of throughput.

Application Bandwidth/User Notes
Text messaging / IM < 1 Kbps As traffic is infrequent and asynchronous, IM will tolerate high latency.
Email 1 to 100 Kbps As with IM, email is asynchronous and intermittent, so it will tolerate latency. Large attachments, viruses, and spam significantly add to bandwidth usage. Note that web email services (such as Yahoo or Hot-mail) should be considered as web browsing, not as email.
Web browsing 50 - 100+ Kbps Web browsers only use the network when data is requested. Communication is asynchronous, so a fair amount of lag can be tolerated. As web browsers request more data (large images, long downloads, etc.) bandwidth usage will go up significantly.
Streaming audio 96 - 160 Kbps Each user of a streaming audio service will use a constant amount of relatively large bandwidth for as long as it plays. It can tolerate some transient latency by using large buffers on the client. But extended periods of lag will cause audio “skips” or outright session failures.
Voice over IP (VoIP) 24 - 100+ Kbps As with streaming audio, VoIP commits a constant amount of bandwidth to each user for the duration of the call. But with VoIP, the bandwidth is used roughly equally in both directions. Latency on a VoIP connection is immediate and annoying to users. Lag greater than a few milliseconds is unacceptable for VoIP.
Streaming video 64 - 200+ Kbps As with streaming audio, some intermittent latency is avoided by using buffers on the client. Streaming video requires high throughput and low latency to work properly.
Peer-to-peer filesharing applications (BitTorrent, KaZaA, Gnutella, eDonkey, etc.) 0 - infinite Mbps While peer to peer applications will tolerate any amount of latency, they tend to use up all available throughput by transmitting data to as many clients as possible, as quickly as possible. Use of these applications will cause latency and throughput problems for all other network users unless you use careful bandwidth shaping.

To estimate the necessary throughput you will need for your network, multiply the expected number of users by the sort of application they will likely use. For example, 50 users who are chiefly browsing the web will likely consume 2.5 to 5Mbps or more of throughput at peak times, and will tolerate some latency. On the other hand, 50 simultaneous VoIP users would require 5Mbps or more of throughput in both directions with absolutely no latency. Since 802.11g wireless equipment is half duplex (that is, it only transmits or receives, never both at once) you should accordingly double the required throughput, for a total of 10Mbps. Your wireless links must provide that capacity every second, or conversations will lag.

Since all of your users are unlikely to use the connection at precisely the same moment, it is common practice to oversubscribe available throughput by some factor (that is, allow more users than the maximum available bandwidth can support). Oversubscribing by a factor of 2 to 5 is quite common. In all likelihood, you will oversubscribe by some amount when building your network infrastructure. By carefully monitoring throughput throughout your network, you will be able to plan when to upgrade various parts of the network, and how much additional resources will be needed.

Expect that no matter how much capacity you supply, your users will eventually find applications that will use it all. As we'll see at the end of this chapter, using bandwidth shaping techniques can help mitigate some latency problems. By using bandwidth shaping, web caching, and other techniques, you can significantly reduce latency and increase overall network throughput.

To get a feeling for the lag felt on very slow connections, the ICTP has put together a bandwidth simulator. It will simultaneously download a web page at full speed and at a reduced rate that you choose. This demonstration gives you an immediate understanding of how low throughput and high latency reduce the usefulness of the Internet as a communications tool. It is available at wireless.ictp.trieste.it/simulator/

Last Update: 2007-01-24