Netgear's WG302 ProSafe 802.11g

Efficiency by Design

By automatically configuring and coordinating channel allocation of each AutoCell-enabled AP in your infrastructure, as well as continuously adjusting power output levels so as not to stomp on neighboring APs, Propagate's AutoCell takes much of the tedium out of the WLAN cell setup.

Client cards supporting the technology will realize some benefits, too. For instance, with pre-emptive intelligent roaming, where client cards look to connect to other APs before their current connections degrade too much, and load-balancing capabilities, clients will roam more efficiently. Cards that run AutoCell code, such as Netgear's WAG511, adjust their power settings automatically based on their proximity to the APs with which they are communicating, thus limiting interference with nearby APs and clients on the same channels.

I configured IP addressing for three WG302s over their console ports and then logged on to the SSL-encrypted Web configuration pages.To fully test the WG302's RF features, I used AutoCell-enabled Netgear WAG511 802.11a/b/g NICs and Propagate's AutoCell View monitoring and visualization software. After configuring the APs with static IP addresses on the same subnet, I pointed View to that subnet to scan for APs. It identified all AutoCell-capable Netgear APs, the clients associated to them and all other APs within earshot of the WG302s, and represented them visually as movable icons that I could click and drag to their corresponding locations on an input floor map of my test facility.

You'll know where to indicate the AutoCell APs' locations within View, but the locations of other APs, as well as the mobile clients associated to the WG302s, are much harder to pin down. Propagate says a more granular location capability is in the works.In small environments, the WG302's automatic RF features ease the administrative burden. But Propagate's need to touch the client for the WLAN to realize its full potential and the system's lack of more robust management features and advanced RF tuning capabilities make it less than ideal for large organizations.

Read about more details on my tests of the WG302.

Jesse Lindeman is the lab manager at the Center for Emerging Network Technologies at Syracuse University. Write to him at [email protected].

To test roam times both with and without AutoCell capabilities enabled, I used an Azimuth Systems 800W and Mini RF Test Head to facilitate a controlled client roam and compare packet time stamps to determine effective roam times. I used the Azimuth testing hardware to isolate the radio waves of each AP and then had one client roam from one AP to another while capturing the transmitted traffic with WildPackets' AiroPeek NX 2.02 packet-analysis tool.

The time from the last transmission on the first AP to the first transmission on the second AP constitutes "roam time." The roaming time for the Netgear WAG511 client card from one WG302 AP to another--all without AutoCell enabled--consistently took about 40 seconds. This is abysmal considering that not only latency-sensitive traffic, such as voice, but all other types of data will be disrupted as the client scrambles to find and connect to another AP.Roaming results markedly improved when I enabled AutoCell on both the AP and client. With AutoCell running, I was able to get the Netgear WAG511 card to roam from one AP to another consistently in about a tenth of a second. All roaming testing was performed with both APs on the same subnet; Netgear does not have a solution for inter-subnet roaming.

I set up four 1.2-GHz Windows XP laptops and bombarded them with 100 iterations of a 1-MB TCP-based, unidirectional file transfer to test throughput capabilities. The WG302 faired well in basic throughput testing for 11b and 11g clients, but slacked in mixed mode testing. Tests for both one and two 802.11b clients consistently grabbed more than 6 Mbps of average throughput. Both single-client and dual-client 802.11g testing showed the AP could consistently reach average speeds of just above 21 Mbps.

Tests with both 11b and g clients wrote a slower story. The AP reached average speeds of about 7.5 Mbps, with clients of both grabbing about equal shares of the pie. This is more a limitation of 802.11g and the way it handles backward compatibility with the slower, but ubiquitous 11b, than it is a problem with Netgear's AP--other companies APs typically grab around 8 Mbps to 12 Mbps of aggregate throughput in mixed 11b and 11g environments, as well.

A proprietary turbo mode is supported that promises to double raw data rates to 108 Mbps, but I didn't test this mode. Selecting this mode eliminates the AP's ability to use the AutoCell features and narrows channel selection from three nonoverlapping channels to just one, making deployments of more than one AP nigh impossible.

To test range, I placed the WG302 in a central, elevated location and performed a basic ping test to determine maximum AP range for both 802.11b and 802.11g. Results were almost identical--roughly 85 percent of the best range results we typically see in our environment. If coverage is more important than capacity, range can be extended using an add-on power booster or any of a number of available Netgear high-gain omni- and semi-directional external antennae.The View From Here

Basic AP configuration information is available through View. From there I verified the automatic channel and power output selection of the WG302. At power on, each AP chooses an optimal channel among 1, 6 and 11 (in the United States). From then on, channel allocation doesn't change--a potential problem if noise sources crop up. Power output settings change, however, to complement the coverage areas of other AutoCell APs. A spectrum analyzer and AirMagnet's Trio WLAN analysis tool helped me verify the APs' selections made sense considering our lab's harsh 2.4-GHz environment.

I also verified that clients roamed to neighboring WG302 APs with fewer and faster client connections looking for more bandwidth on less congested APs. Unfortunately, actual AP utilization isn't taken into account, so there is the likelihood that clients may make ill-informed roaming decisions on occasion.

Some Gotchas

Noticeably absent is 802.11a support. Adding 802.11a capabilities would please more demanding wireless admin, because of the standard's higher effective throughputs and the fact that there are fewer interference issues in the 5-GHz bands and more non-overlapping channels available--currently 12, with spectrum recently opened to allow a total of 23. If Netgear wants to compete, it must add 11a.Although the user guide indicates initial setup should be performed by logging in from a network PC directly to the AP's factory-set default static IP address, configuring the WG302 this way can be a real hassle with multiple APs. Potential addressing conflicts would require they be powered on in succession as each is manually handed a new, non-conflicting static address or is set to get one dynamically. Initial configuration would be greatly simplified if Netgear provides a basic configuration utility to make setting up multiple APs easier or, at very least, if it had set its APs to grab DHCP addresses automatically.

Security-wise, the WG302 offers what's expected; it's got support for both 802.11i and WPA--a WiFi Alliance-branded subset of the newly ratified security standard.

The WG302 is manageable through its SSL-secured Web page and SNMP, but, putting it above other similar consumer-grade offerings, it also allows admin to manage via SSH and console port, as well as through AutoCell View.