How to get backward compatibility in next gen WLANs

How to get backward compatibility in next gen WLANs
Designers of next-generation high-speed wireless LAN (WLAN) devices are primarily concerned with two issues: Which high-speed standard, 802.11a or the proposed 802.11g, will offer the best solution? And, how do they address backward compatibility with a legacy 802.11b infrastructure?

Designers can choose from two major architectures: the dual-protocol stack (DPS) architecture, which is frequency-agnostic, and the common-air interface (CAI) architecture, which is frequency-centric. Each approach requires design tradeoffs for developing the optimal solution for a WLAN application.

A powerful way to compare the 5-GHz 802.11a standard with the proposed 2.4-GHz 802.11g standard is to consider the major features and specifications: system capacity, total system throughput, system interference, and interoperability with legacy 802.11b products.

See related chart

When evaluated on these major performance issues, the ratified 802.11a standard clearly has an advantage over the proposed 802.11g standard. In terms of time-to-market, the 802.11g specification is still in the early stages of the ratification process, and products aren't expected until mid to late 2003.

Even though it is considered an extension of 802.11b, it is important to consider 802.11g interoperability and interference issues with legacy 802.11b equipment. The current draft 802.11g standard does not clearly define interoperability with 802.11b-only products, causing valid concern among designers considering the 802.11g approach.

As new WLAN equipment arrives in incumbent installations, backward compatibility will become very important. This analysis of backward compatibility will focus on the ability of products to offer a dual-mode system that simultaneously supports the chosen high-speed standard (802.11a or 802.11g) and the established 802.11b standard.

For some time 802.11a and, eventually, 802.11g, and legacy 802.11b products will need to coexist. Network administrators will most concerned with deploying backward-compatible high-performance products into existing wireless networks to ensure a seamless upgrade path to a higher performance WLAN. Those who do not currently have a WLAN system installed will likely move directly to implementing a high-speed system.

Since 802.11a and 802.11b operate in different frequency bands, they do not cause performance degradation. The same is not known about the proposed 802.11g products, which operate in the same 2.4 GHz band as 802.11b and numerous other products, such as Bluetooth-enabled devices. In addition, there have been no definitive studies on how legacy 802.11b equipment will handle the energy produced by the new high-data-rate modes of 802.11g. At a minimum, there is likely to be severe degradation in the performance of the legacy network.

Before deciding on a design, it may be helpful to consider the issues from the perspectives of the access point (AP) and the client. New APs are being designed for dual-mode simultaneous operation. Typically, the system architectures accommodate dual network-interface card (NIC) slots — one for a factory-installed 802.11b and another for future expansion using a single-mode 802.11a or other NIC card to create a dual-mode AP.

From the client perspective, designers can safely assume that legacy networks are 802.11b-based, and that new high-speed networks will encompass high-performance, high-throughput products.

WLAN designers must also ensure that clients do not lose their service while roaming between APs, whether using 802.11b or another high-performance standard such as 802.11a or the proposed 802.11g standard.

See related chart

After understanding these standards and compatibility issues, the designer will need to select an architecture. Two options for developing backward-compatible systems in 802.11a and proposed 802.11g standards-based products include the dual-protocol stack and the common-air interface approach.

DPS has two distinct physical layers that operate simultaneously. A physical realization may be a single MAC+baseband processor (BBP) chip with two external radios, or a single MAC+PHY chip that has an external bus interface to an additional separate PHY chip, or a single MAC chip that has a bus interface to two separate PHY chips.

Containing a single MAC and separate PHY for each protocol, the dual-protocol stack architecture is well suited for use in backward-compatible 802.11 b/a solutions. Because it is frequency-agnostic, it easily supports simultaneous operation in both the 2.4 GHz and the 5 GHz bands. However, this positive attribute becomes an issue when applied to an 802.11b/g solution, which operates only in the 2.4 GHz band. In this case, there are potential implications from a network planning perspective due to the limited number of channels available.

A potential drawback to the DPS approach is that it requires two radios. This possible design negative can be offset by using two optimized low-power PHYs, a feature that provides a superior high-performance solution. Support for simultaneous and non-simultaneous dual-mode operation makes the DPS architecture suitable for both AP and client devices.

The common-air interface is another approach to designing backward-compatible solutions. This architecture defines a physical layer structure that allows both protocols to exist on the same channel and share the same physical medium. The various protocols, which may use different modulation schemes, time-share the channel. The physical realization may be either a single MAC/PHY chip, or a single MAC chip that has a bus interface to a single PHY chip.

Because the CAI time-shares one channel to provide backward compatibility, it is better suited for use in 802.11b/g backward-compatible solutions. However, it is not an optimal choice for 802.11b/a backward-compatible solutions because it does not allow for simultaneous operation in two different bands.

The time-sharing aspect of the CAI architecture increases the difficulty of supporting simultaneous dual-mode operation, limiting the attractiveness of its use in AP applications. In addition, time-sharing a channel in 802.11b/g solutions reduces throughput, exasperating the problem of limited channel capacity and impacting network planning.

This architecture uses only one radio; however, it is a traditional radio architecture that consumes a significant amount of power when used in orthogonal frequency division multiplexing (OFDM) transmission. The CAI structure still requires two baseband processors, and the associated incurred costs, to practically implement dynamic switching between modes within the timing constraints of each protocol.

The DPS architecture is suitable for both access points and client applications. However, the limitations in system performance such as system throughput and channel capacity, and network planning complications of 802.11b/g DPS architectures make the 802.11b/a DPS an optimal architecture for backward-compatible high-speed WLAN APs. The DPS architecture's requirement for a second radio can be offset by the use of two optimized low-power, high-performance PHYs to improve mobility and performance.

Overall, designers looking to implement a next-generation solution in AP or client configurations will benefit from selecting a ratified high-speed standard, 802.11a, and implementing it in a DPS architecture.