Embedded Wireless Made Easy

GARY MARRS
Senior Field Application Engineer
Lantronix Inc.
Irvine, Calif.

An industrial manufacturer recently took its first stab at building wireless capabilities into one of its products. Engineers there spent two years trying to devise an RF interface that would pass FCC certification tests. They went through several redesigns and chewed up about a million dollars in the process. After the second time their efforts failed the FCC hurdle, they decided to throw in the towel and buy an off-the-shelf WLAN module instead.

The experience of this manufacturer is not uncommon. The design of RF circuitry is still a discipline that is part art and science. It can trip up the uninitiated who underestimate the resources necessary to field equipment that stays within strict guidelines spelled out by regulatory agencies.

The difficulty of this task is one reason for the advent of standard modules that can be embedded into existing products to serve as self-contained WiFi Web servers or nodes on a wireless LAN (WLAN). These modules are designed to work as coprocessors that handle wireless network communications. They offload the host processor by handling all WLAN tasks and provide a complete drop-in subsystem for firms adding a WLAN interface to their product.

Before any product with an RF transceiver can be fielded, it must be approved by the appropriate agency that regulates intentional radiated noise. In the United States, this is the FCC. FCC approval requires extensive testing and the testing can be costly. It must take place at a certified test lab. It is not uncommon for such labs to charge upwards of $400/hr for time in the anechoic chambers where they conduct radiation and emission tests. Total costs of $50,000 for FCC compliance testing are not uncommon. One reason the price tag is so high is that designs rarely make it through FCC testing on the first pass.

By using a standard WLAN module, manufacturers can in many cases eliminate the need for testing to confirm FCC compliance. Module makers get certification for their devices and this certification typically covers use with any antennas having a gain of 3 dBi or less. That includes widely used models such as 'rubber ducky' antennas. That means OEMs incorporating these modules avoid the expense of compliance tests as long as the antennas they deploy fall within these general areas.

Designs that use self-contained WiFi modules avoid the need for FCC testing with one important exception: Modules that use antennas with a gain exceeding 3 dBi must go through the same FCC compliance tests required of designs that are built from scratch. However, high-gain antennas are not usually required but come into play when it is necessary to get a longer range than 100 m or for higher link margins in noisy environments.

Use of a standard module conserves resources in other ways besides FCC testing. Most WLAN modules take serial data to and from the network. If the current product already has a serial (RS232) interface, addition of the WLAN module generally takes place with no changes to the host processor firmware. This saves hundreds of hours of engineering time and allows companies to roll out wireless capabilities in six months to a year less than the time needed otherwise.

The 11s: A, B, And G
IEEE 802.11 is the protocol that serves as the primary standard for wireless network and Internet access. There are three main versions, the most widely used being 802.11b. The 11g version is a recent enhancement of 11b.

Products built to the 802.11a standard are much less widely used than those following 11b. Products for the 802.11a standard actually appeared after those for 11b. 11a uses the 5-GHz band; 11b and 11g use the 2.4-GHz band.

This makes 11a products completely incompatible with the other two standards. The use of the higher frequency gives 11a the ability to transmit at a higher bandwidth, yielding a throughput of up to 54 Mbytes/sec. This is fast enough to manage tasks such as transmitting DVD-quality video, something impractical with 11b.

On the other hand, equipment for 11a tends to be more costly than that for 11b because the higher band necessitates use of components that are more expensive. The range of 11a systems also tends to be shorter than that of systems operating at the lower frequencies due to higher path losses. These drawbacks have kept 11a gear out of most machine-to-machine type uses. However, some OEMs are moving to this higher frequency to attain higher data rates and less overcrowding.

The most widely deployed wireless standard today, particularly for industrial use, is 802.11b. It has a raw data rate of 11 Mbytes/sec but, because of the overhead of its CSMA/CD protocol, maximum throughput is about 5.9 to 7.1 Mbyte/sec depending on whether TCP or the simpler UDP transmission protocol is used. 802.11b usually gets deployed in a point-tomultipoint mode where an access point communicates through an omni-directional antenna with one or more other nodes. 802.11b equipment will automatically scale back from 11 Mbyte/sec rates if signal quality degrades. This is because lower data rates use simpler and more redundant methods of encoding data, so they are less susceptible to interference and corruption.

Finally, the 802.11g specification is a faster and recently released upgrade of 11b that uses the same transmission frequencies. It provides data rates to 54 Mbyte/sec and is backwards compatible with 11b transmissions.

Today, most 11g applications are in the IT area. And 11g equipment is slightly more expensive than that for 11b. Nevertheless, this cost difference will eventually disappear and 11g will likely become the next widely used standard for wireless nets.

Secure Links
No question that security is a hot topic, and the security of wireless communications in particular have drawn a lot of scrutiny. Actually wireless links are as secure as wired connections. Wireless security involves encryption and authentication algorithms that can make it difficult to intercept and decode the data.

Though there are several encryption techniques used today, two in particular find wide deployment. Wired Equivalent Privacy (WEP) is the most common but also one of the easiest to break. It uses what's called a symmetric key to scramble and unscramble data. Here the sender and the receiver must have the same key to communicate. The problem with WEP is that snoopers can figure out the key if they intercept enough messages. In fact, there are algorithms available on the Internet that let hackers break into WEP connections fairly quickly.

WiFi Protected Access (WPA) is becoming more widely used for applications that need stronger security. WPA frequently jumbles the messages it transmits. It still uses a key, but the constant changes make the task of a snooper more difficult. WPA also builds in the ability to authenticate the network access point. This feature gives WPA nodes enough smarts to detect attempts at impersonating legitimate network participants.

Wireless equipment now coming off the drawing boards generally implements WPA security measures. In fact, it is likely that WEP will all but disappear in the next year or so. But though WPA is good news for wireless security, it tends to lengthen the development cycle for firms intent on designing their own wireless RF interface. This is because the algorithms involved are more complex than those of earlier security measures.

In contrast, security algorithms are not an issue for manufacturers that use third-party WLAN modules. The module itself incorporates all of the security algorithms. Communications from the module to other networked devices are protected by whatever security measures the module is configured for.

Module Resources
A typical WLAN interface module is designed for use in embedded applications. The typical architecture includes a processor for communication tasks, onboard memory generally made up of flash, high-speed static ram and an RF transceiver front end.

These modules run under the direction of an operating system which, among other things, implements the functions of an embedded Web server, a full TCP/IP protocol stack, and security algorithms. On-board flash stores the embedded firmware application and also lets users store custom Web pages. Firmware and the custom Web pages are upgradeable over the network. Memory resources tend to be small by desktop PC standards, but entirely adequate for most embedded uses. Modules today can be had with up to 4 Mbytes of flash and 1 Mbyte of static ram.

Modules operate basically as a coprocessor that communicates with the host CPU over a serial link. Generally speaking, functions of the module are independent of any operations the host CPU handles. The host CPU treats the module like any other device that sends and requests information over serial lines. Thus there is little software recoding involved in adding wireless capability to most existing applications. Engineering development focuses on connecting the module and adding custom Web pages in applications where the module functions as a Web server.

Data sent between the host CPU and the WLAN module gets encapsulated in TCP or UDP frames. The WLAN module can operate either as a Web server or as a client/server for standard networking schemes. The WLAN Module's Web server can host Web pages developed using standard tools. Custom Web pages, for example, can be written in html and contain JavaScripts. The JavaScript on these pages handles user interface functions while CGI callbacks in the Web server gather data from the host CPU and display it on a Web page. Web pages hosted on these modules get developed with the same Sun Java tools used for devising conventional Web pages.

When evaluating the WLAN module, perhaps the most important criterion is the radio's ability to make a wireless link in noisy environments. With overcrowding of the spectrum and competition from non-WLAN sources such as wireless phones and microwave ovens, the WLAN module needs to have a robust radio. One feature to look at carefully is the module receiver sensitivity. Receiver sensitivity is the minimum signal level for the receiver to be able to decode the data. The lower the receiver sensitivity, the better the radio can decipher good signals from noise. Also, the flexibility to mount the antenna in an optimum location on the device can make a big difference in the quality of the wireless link.

All in all, Wireless is hot technology. Use of an off-the-shelf WLAN module can save an OEM significant amounts of engineering resources and speed time-to-market.