Understanding QoS issues for 802.11

The wireless protocol's popularity has led to some renegade networks. Certain steps must be taken to ensure quality.

There's little question that 802.11 is the wireless technology of choice for enterprise and SOHO data networks. But popularity has forced the standard to undergo a bit of adaptation in an effort to meet the emerging market needs. In the past, 802.11a and 802.11g were developed to meet increased bandwidth requirements. Now, as multimedia and data-centric devices begin to share the same airspace, 802.11 needs to adopt quality of service (QoS).

To accomplish that, the IEEE formed the 802.11e task group to address QoS for 802.11 technologies. As traffic loads increase on a wireless LAN (WLAN), high-priority and real time links lose reliability because nodes contend with each other to determine which one will get to transmit. QoS is essential for enabling the widespread deployment of new wireless multimedia applications such as cordless VoIP phones for enterprise WLANs as well as wireless bridges between analog audio-video equipment and PCs serving up digital content in the home. QoS schemes seek to improve the reliability of high-priority and real time data on the WLAN and reduce contention overhead, thereby increasing the effective bandwidth.

Implementing QoS can create issues, but fortunately there are several ways to address those issues that don't require modifications to the 802.11b/a/g standards. A simple but expensive approach is to overprovision bandwidth. For networks that already provide the maximum bandwidth capacity, there's frequency diversity, which isolates QoS traffic on specific RF channels, effectively dedicating bandwidth to this traffic. However, there are some tradeoffs. First, there's the cost of having duplicate equipment. Then, we must consider how bandwidth is affected for applications that are making use of all the available channels to achieve wider WLAN coverage. Finally, this approach doesn't fundamentally solve the QoS problem, making it an unattractive long-term alternative.

Another more basic approach is to queue traffic in the host, place all QoS traffic in a high-priority queue, and separate it from low-priority data. In this scenario, the host can then preferentially select high-priority traffic for transmission. While this approach improves performance for QoS traffic, it too falls short because the WLAN nodes still compete with each other for access to the medium before they can transmit.


802.11e: two new QoS mechanisms

The default medium access protocol used in 802.11a/b/g is DCF (distributed coordination function). In general, when the host system requests that the WLAN MAC should transmit a frame, the MAC will check the physical and virtual carrier sense mechanisms for an indication that the medium is free for a period of DIFS (distributed interframe space) and that the MAC back-off timer has expired. If these conditions are met, the MAC can transmit. If the wireless medium is busy at the time of attempted transmission, the MAC will defer until the wireless medium is free, then wait for DIFS and a back-off period before attempting transmission.

Contention has an effect of reducing the overall bandwidth because devices must back off before attempting to communicate again. That compounds the difficulty for a real time data transfer to occur before the real time deadline has passed. To reduce the possibility of protracted contention, the back-off period includes a random element. If there weren't a random element to the waiting period, then every device with the same priority would contend again after the waiting period.

A primary advantage of DCF is that it provides an efficient low-latency mechanism for transferring bursty traffic. The chief problem with DCF is that all devices talking to an access point use the same mechanism and thus all receive the same QoS. A real time data source such as voice or video will face continuous contention from a large data transfer happening at the same time. Unfortunately, DCF doesn't provide enough control over how bandwidth is allocated.



EDCA enables QoS by reducing the interframe gap for higher-priority traffic.

The 802.11e draft specification provides two new mechanisms for resolving contention that enables QoS: EDCA (Enhanced Distributed Channel Access) and HCCA (Hybrid Controlled Channel Access). EDCA, also known as prioritized DCF, improves on DCF by giving higher-priority traffic an advantage during contention. Instead of waiting a DIFS period before transmitting after the back-off period expires, higher-priority traffic can attempt to transmit only after a PIFS (point coordination function interframe space) period and associated back-off time (see the figure). Using the EDCA scheme, nodes that offer high-priority traffic, an example being VoIP phones, have a higher probability of gaining channel access than the nodes offering lower-priority traffic. Low-priority traffic might be PCs performing file downloads.

Granularity within priority slots is possible by carrying this technique to exponential back-off. Higher-priority devices will back off at a slower rate (i.e., fewer slot times) than lower-priority devices, giving higher-priority devices another contention advantage. Additionally, EDCA coexists with DCF-based devices because all DCF devices appear as low-priority (DIFS) nodes.

Using EDCA, an access point can support more VoIP phones than DCF, for a given voice quality. However, EDCA doesn't give rock-solid deterministic QoS for every application. Aggressive high-priority nodes can effectively shut out low-priority nodes.

Guaranteeing a minimum bandwidth to every node requires a polled mechanism like HCCA, which also supports the QoS required for more demanding applications such as streaming media. Nodes with priority traffic request to be added to a polling list managed by the access point. The access point in turn polls each QoS node, and in doing so, avoids contention with DCF and EDCA nodes by transmitting polls before any of these devices can begin to contend for channel access.

HCCA is a hybrid mechanism, merging the best aspects of a coordinated mechanism for QoS traffic and DCF for efficient handling of bursty traffic. Since HCCA preempts the network for QoS traffic, it must regularly go quiet to allow non-HCCA-enabled devices to transmit as usual. During these quiet times, EDCA-based devices can gain channel access, as can DCF devices, and low-priority data traffic can efficiently squeeze in between QoS frames. That is an important aspect of HCCA because it's likely that networks will contain a mix of HCCA-, EDCA-, and DCF-based devices that must coexist with each other.


What about WME?

Quality of service is such an important enabling function that many companies would like to implement QoS into their products immediately. EDCA and HCCA as defined in the draft 802.11e specification are still subject to final approval within the IEEE standards process. Complicating the issue is that, for many applications, EDCA provides enough QoS functionality. Vendors wanting to introduce QoS-based products may see EDCA as an attractive alternative because it fulfills near-term needs for QoS functions and is easier to implement than HCCA. The problem is that early implementations of EDCA risk being proprietary because the 802.11e standard isn't approved.



The WiFi Alliance (WFA) has developed wireless multimedia enhancements (WME) in an effort to discourage the market fragmentation that comes with multiple proprietary QoS mechanisms, while meeting the needs of companies that want to start QoS implementation now. It's important to note that WME is not an alternative to 802.11e. Rather, WME is a subset of the 802.11e standard, based on EDCA wrapped with enough bits and pieces to create a complete and self-consistent QoS toolkit (see the table). In other words, WME is a snapshot of the 802.11e standard, developed in an amicable manner with the IEEE. WME has been developed to comprehend forward compatibility with 802.11e systems. While it doesn't meet every QoS need for certain applications, it does provide an intermediate step to increase bandwidth efficiency that may carry such applications until HCCA in 802.11e is approved.

The Wi-Fi Alliance is also developing WSM (wireless scheduled multimedia), HCCA's counterpart to WME. WSM will provide a platform for testing compliance and interoperability for both HCCA and EDCA devices for Wi-Fi certification. WME plugfests are underway, and expectations are that implementation details and the Wi-Fi certification process will be sorted out very soon. The 802.11e spec is expected to be approved by mid-year.

Once 802.11e is ratified, many WME devices will be upgraded to 802.11e with HCCA through software upgrades. Designs based on off-the-shelf silicon will be able to roll out software upgrades available from the silicon vendor to incorporate full EDCA and HCCA, along with optional features (see the sidebar "Above and beyond quality of service," page 17), as well as take advantage of new reference designs incorporating these features, such as access points, PC cards, and VoIP phone modules.

But in the same way that not all 802.11 transceivers support the AES security encryption standard merely with a software upgrade, not all of them will support HCCA with existing silicon. With AES, the transceiver needed access to a hardware encryption engine that could encrypt/decrypt data in real time. Without such an acceleration engine, the transceiver couldn't run at full rate, unless a heavy processing load was imposed on the host CPU.

For 802.11e, the issue is how fast the hardware can turn around from receive to transmit modes. Existing architectures were designed to turn around in a DIFS period, or 50 µs. To support higher traffic priorities, the architecture must be able to turn around in a short interframe space (SIFS) period (10 µs). To achieve such low latency requires a double-buffered architecture enabling a frame to send when it's time to transmit. If the hardware can't initiate transmission within a SIFS period, then even with a software upgrade the device may not be able to support the highest service priority.

Given that there are applications for which EDCA provides sufficient QoS functionality and applications that require a more controlled mechanism like HCCA, silicon for access points will need to support both mechanisms. High-priority nodes may want to support both mechanisms as well so they can provide the best QoS for both EDCA-only and HCCA-based access points.

Again, WME isn't an alternative to 802.11e. The introduction of WME is anticipated to provide at least a six-month advanced introduction of EDCA-based devices. Not only will it help maintain WiFi interoperability, WME also will actually accelerate the release of 802.11e devices, because development teams will have had a chance to implement and certify EDCA devices earlier than if they had to wait for the entire 802.11e spec's ratification.

Ian Sherlock is a product manager for Texas Instruments' WLAN product line. Based in Dallas, the company can be reached at www.ti.com.


Bluetooth and 802.11b

It may be suggested that Bluetooth competes with 802.11b; though they're similar (e.g., they occupy the same 2.4-MHz frequency band), Bluetooth is intended to be a wireless "cable and infrared replacement," while 802.11b is suited for wireless LANs.

When considering which technology best suits a specific need, users must choose which feature set is most appropriate for a given application. That depends on a number of factors, including the volume of data to be exchanged; range of data transmission; and available power from the host device (battery-operated systems have limitations due to the greater power required for 802.11b implementations).

Bluetooth has a maximum range of 100 m (10 m nominal) at a 721-kbit/s data rate, while 802.11b has a range of more than 100 m with a data rate of 11 Mbits/s. Based on this criteria, Bluetooth's features are best suited for synchronizing data between PDAs, PCs, and cell phones, while 802.11b offers a full Ethernet-style network.

Market data indicates that 802.11b has become the dominant WLAN technology in the market, which may indicate more general availability from an "end user" perspective (it's considered a more "seasoned" technology), and would appear to be the more pervasive technology. However, a wider range of personal-area-network (PAN) applications and products have entered the mainstream markets, such as headsets for hands-free cell phones, printers, and PDAs.

With the growing availability and emerging business model for Wi-Fi hot-spots compounded by the new surge of products to facilitate a wireless environment, inevitably the latest "hot topic" is now on the security surrounding the information being passed between devices. When transmitting personal information over Bluetooth, there's a level of security that hasn't been found when using 802.11b.

Bluetooth allows a device to be "locked down," which removes some of the availability of that information to anyone trying to hack the device. When using 802.11b, though the issue of security is currently being addressed, wireless channels occasionally drop data packets, so a motivated hacker may be able to pull the key code (which is used to keep the information secure) from that information and gain access to all the transferred data. While users want the speed and distance offered by 802.11b, Bluetooth is ideal for pairing PAN devices that would lend themselves in the transmission of more personal information with a need for a level of security that may not be available with an 802.11b connection.

Ross Forman , Smart Modular Technologies
      Portable Design March, 2004
      Author (s) : Ian Sherlock

    【Print】  【Mailto】