The current limitations of 802.11 networks have held back many consumer and enterprise applications such as video and VoIP. Some of the often heard complaints of 802.11 networks are signal range, security, lack of a quality-of-service (QoS), and maximum data throughput. Because wireless bandwidth is restricted due to the potential interference in the already crowded 2.4 GHz range, QoS is becoming increasingly important in 802.11 networks.

QoS and security were supposed to be addressed in the next 802.11 wireless protocol standard, 802.11e. However, the security issues took a life of its own and are now split off into a separate protocol 802.11i thus only leaving QoS enhancements in 802.11e. Uncoupled from the security requirements, 802.11e is expected to get ratified sometime in Q3 2004. The sooner this new standard arrives, the better. Wireless users are starting to demand applications that wired users are afforded, such as streaming video and sending large data files.

Lack of a QoS is due in today's wifi standards was part to the original 802.11 media access control (MAC) protocol. The MAC protocol was designed with two modes of communication; distributed coordination function (DCF) and the less widely used point coordination function (PCF). DCF is based on carrier sense multiple access with collision avoidance (CSMA/CA), which basically means listening before talking. While DCF will detect collisions on the network, it doesn't offer any network access prioritization. PCF supports time-sensitive traffic flows. PCF splits time into a contention-free period and a contention period. Using PCF, a client can transmit data only during contention-free polling periods.

With 802.11e, enhancements are being made to both modes to aid QoS. These enhancements will be backwards compatible with existing 802.11 networks while addressing QoS issues. The enhancement to DCF in 802.11e is the enhanced distribution coordination function (EDCF), which introduces the idea of eight priority levels. Under EDCF mode, a client will try to send data when the network is contention-free and after waiting a period of time defined by the arbitration interframe space (AIFS). Stations with higher priority traffic such as VoIP and video will have a shorter AIFS and therefore waiting time than client with web browsing traffic.

All this talk of QoS might lead one to think it is the holy grail of Wi-Fi. QoS technologies are designed to guarantee timely delivery of specific application data or resources to a particular destinations. Some of QoS's advantages are guaranteed bandwidth for key applications and users. It can also delay a costly upgrade to a faster network infrastructure and help in network planning by measuring and managing traffic flow.

What does this all mean for us? Earlier today, Agere Systems announced performance acceleration software for its WaveLAN 802.11a/b/g multimode chipset. This software is capable of achieving up to 150 Mbits throughput for wireless products. This "turbo mode" capability was developed using 802.11e's QoS techniques and other software enhancements.

Some of the techniques used to improve data transmission speeds are direct link protocol (DLP), a mechanism to allow two devices, if they are in range of each other, to bypass the access point and send data directly to each other. Contention-free bursts (CFBs), used to improve efficiency by eliminating some contention. When an access point has time remaining to transmit in a granted transmit opportunity (TXOP) and additional data to send, instead of contending for the medium again, 802.11e allows a station to resume transmitting after a short delay. Additional techniques include data compression and overhead management.

Ramping to production this month, Agere's WaveLAN multimode chip set consists of a single-chip, dual-band 2.4/5.2 GHz RF transceiver; media access controller (MAC), baseband processor and power amplifier. The WaveLAN chip set is expected to ship in the third quarter of 2004 along with Agere's acceleration software which will be implemented as a technology overlay. Unless there is unforeseen deadlock in the standards committee this should coincide with the expected ratification of the 802.11e standard. This market timing will ensure that Agere's software enhancements fully comply with the final QoS specification.

Although this all sounds good on paper, it is not until the product ships and additional real world tests are performed before we can say Wi-Fi might be able to replace wired connections for high throughput demanding users. Looking at the Agere's announcement one has to wonder if utilizing 802.11e's QoS, DLP, and CFBs features along with compression is enough to almost triple the data throughput? Unfortunately, Agere's claim of twice the range of rival chipsets at peak data rates is a little ambiguous. Numerous websites, including Agere's own site, just offer vague references. There has not been any explicit mention of actual distances.

Although late to the wifi party, Agere's stance was to wait until it had a dual-band solution that would cover all three popular WLAN standards. If Agere is able to get real world performance near 100 Mbits with a good QoS they might have a killer chipset. Also, if they are able to do this up to twice the range of other 802.11e chipsets, we would have a new dominant chipset to contend with. Some of the companies slated to use this new chipset are Samsung, Ubicom, Accton, Ambit Microsystems, CyberTAN Technology and Universal Scientific Industries. Hopefully, with a lower bill of materials associated Agere's chipset, products that utilizes this chipset will have a price point comparable if not lower than other 802.11a/b/g APs. We will find out late this summer when products with Agere's chip set and software ships.

Comments Locked


View All Comments

  • amdfanboy - Wednesday, March 17, 2004 - link

    All I have to say is standards. If you don't find ones that suit you, make your own. Brian should come visit the quiet and cozy network forum.
  • KristopherKubicki - Wednesday, March 17, 2004 - link

    Agere has been working on this for a *long* time - i remember talking to them at COMDEX and they were still working on the chip then. Agere is a lot of ex-AMD folks- i dunno - i say good luck to them.

  • jhevenor - Wednesday, March 17, 2004 - link

    This notion of priority that is mentioned is bogus for two reasons.

    First, do these high priority stations pay more for their shorter wait times? And once their data is in the rest of the internet does it get any special treatment? No.

    Second, if they do get a priority system in place for these applcations then what is to stop all applications from making their data high priority. A wifi router is not going to pattern match each packet to ensure it's of a specific application type. Not without significant delay.

    So in short, I don't buy it.
  • MadAd - Wednesday, March 17, 2004 - link

    Actually, while im thinking about it, theres one thing that I rarely see mentioned in all these wi-fi reviews - latency. Does noone game on wi-fi?
  • MadAd - Wednesday, March 17, 2004 - link

    Kris in 'denys all' shocker :)
  • KristopherKubicki - Tuesday, March 16, 2004 - link

    These networking articles are actually written by a new contributing writer- Brian Ng. He is set to start doing networking reviews for AT. Its just getting published under my byline until he gets his feet wet.

  • TrogdorJW - Tuesday, March 16, 2004 - link

    Wow! Look at all the comments! :)

    It's not that we don't care, Kristopher... it's just that WiFi is not as "great" as so many people seem to think for the enthusiasts. At least, that's my take.

    I look at the throughput graphs for 802.11A and G and compare that to the speeds I get with 100 Mbit Ethernet, and I can't see any reason to think about wireless right now. I transfer large files between my home PCs regularly, and I max out at around 9.5 MB/s, which is still three or four times as fast as the best WiFi connections. Couple that with the fact that I don't have to worry about conflicting networks in the neighborhood and near-complete security behind a firewall, and I really am not that excited for WiFi.

    Needless to say, if the past is any indication, Agere's claims of "150 Mbps" speeds and "twice the range of competitors" will end up equaling 30 Mbps realistic transfer rates (5 Mbps higher than the current crop) at ranges slightly less than the competition. And they'll probably forget to enable WPA or whatever. :p

Log in

Don't have an account? Sign up now