TGn Sync Improves Throughput and Applications for 802.11
by Rick Bahr, VP of Engineering; Won-Joon Choi, PhD; Huanchun Ye, PhD
Atheros Communications

In Lee Goldberg's Editorial, WAN Wars: The Summer's Hottest Spectator Sport, he invited competing standards' factions for future 802.11 applications to put forward their positions for readers to compare. First up is the following OpEd piece positioning the TGn Sync group. Jim Zyren, representing WWiSE will present his views as soon as possible; unfortunately his input has been delayed while he assists in cleanup after Hurricane Frances.

TGn Sync is a multi-industry group working together to ensure that the next generation of wireless technologies is both faster and better suited for the expanding range of 802.11 applications. The group consists of over a dozen companies from across the cellular, computing, consumer electronics, enterprise, public access, and semiconductor markets based in North America, the European Union, China, Japan, and South Korea. The focus is a new class of low-power, adaptive radio technologies that deliver data rates up to 315 Mbit/s reliably in standard two-antenna designs, and are extensible to over 600 Mbit/s. The TGn Sync proposal offers a scalable framework for managing the planned expansion of both unlicensed and licensed spectrum and its use in 10 MHz, 20 MHz, and 40 MHz configurations. More information about TGn Sync and its comprehensive MAC and PHY proposal are available at http://www.tgnsync.org.

The TGn Sync MAC proposal is based on extending 802.11e draft, now in the IEEE sponsor ballot stage and a part of Wi-Fi Alliance WMM (Wireless Multimedia) testing. This proposal enables strong support of legacy 802.11 products along with significant enhancements to take advantage of the increased data rates of the proposed MIMO (Multiple Input Multiple Out) OFDM PHY, as well as to expand its suitability for different usages and classes of devices. This means, for example, that ultra-low-power modes are supported for small, battery-powered devices such as cellular phones.

Below are some of the more salient features:

The ability to reliably support high data throughput over range is a key goal of the TGn Sync proposal. Compared to other proposals, the TGn Sync design provides a more flexible and robust frame aggregation structure, with Cyclic Redundancy Checks (CRC) on member MAC Service Data Units (MSDU), as well as MSDU delimiters. This CRC localization more efficiently manages different usage profiles by preventing an error in one MSDU from propagating into the next MSDU. In this way, also, only the MSDU in error need be retransmitted, and the frame error rate becomes independent of the length of the aggregate. This gives the transmitter freedom to optimize the frame and aggregate sizes for best performance according to the link condition. The aggregated frame can be addressed to single or multiple receivers, which helps maintain MAC efficiency for applications with (for instance) significant voice traffic. In addition the TGn Sync MAC provides support for closed-loop MIMO channel estimation that is used in beamforming, and channel quality estimation that is used in improved rate adaptation for best PHY performance.

Traditionally, high-speed designs are at odds with, and poorly suited to, either (1) low-power devices or (2) devices that need less throughput more reliably. The TGn Sync MAC provides protocol-level support to enable clients to reduce power consumption, with an eye toward battery-powered mobile stations. For example, when an aggregated frame is addressed to multiple receivers, the list of receiver addresses will appear at the beginning of the aggregate frame, so that stations not on the list can enter a very low-power state for the rest of the frame. The proposal also introduces a scheduled response scheme, whereby a station can sleep until its response is due, further reducing its power consumption. For devices more interested in QoS (Quality of Service) than maximum throughput, the proposal provides a mechanism for reverse-direction data flow, which allows stations to request, and the AP to grant, transmission opportunities, thereby enhancing support for consistent performance.

With the dramatic expansion of both unlicensed and licensed spectrum since 802.11's inception in 1997, it would be a startling oversight to not take better advantage of the frequencies available. The TGn Sync MAC provides a comprehensive method for the co-existence of both legacy and new products that use spectrum in different ways: for example, 20 MHz and 40 MHz, higher speed and legacy speeds. These are in addition to the legacy coexistence mechanisms built into the PHY, and are focused on how to support new and older products as fairly as possible. With the success so far of 802.11, it'll be a long time before pre-802.11n (802.11a,b,g) products disappear from the market. New protocols need to perform well in the mixed use case, and not just greenfield opportunities.

The TGn Sync PHY proposal is focused on using every tool available to both increase throughput and increase the reliability of throughput. This includes the use of multiple antenna technologies such as spatial multiplexing, as well as adaptive radio techniques to manage spectrum, advanced coding capabilities, closed loop feedback, and high precision guard intervals. No single technology acts as the key to the next generation of 802.11n wireless. It's the package of everything that, combined, enables both faster and more scalable 802.11 applications.

Below is a brief overview:

Spatial multiplexing is a technique to achieve higher data rates for a particular signal bandwidth. It uses multiple antennas, but more significantly multiple RF chains at both transmitter and receiver, which mandates increased system costs for additional Tx and Rx components. The TGn Sync PHY proposal provides a full set of capabilities for 2 antenna (RF chain) systems, thereby enabling both today's designs and the more compact devices of tomorrow's markets. The design also supports up to 4 RF chains for high-end applications.

Adaptive radio is a method that increases both the reliability and throughput of a signal at minimal cost by taking advantage of the fundamental tenets of Shannon's law. The spectrum map today is very different than it was in 1997, with allocations throughout the world suited for 10 MHz, 20 MHz, and 40 MHz channelizations. Both simulations and real-world measurements clearly show that the similar high throughput can be achieved at much lower SNR with 2 antennas in 40 MHz than even with 4 antennas in 20 MHz. This led to a design with strong support for adaptive radio capabilities, which means that the resultant protocol is more reliable, and system costs can be greatly reduced

Legacy compatibility is paramount to designs that expect to work well with the large numbers of devices on the market today. A critical part of the TGn Sync design is a robust preamble that is carefully designed to support both compatibility with legacy devices, and interoperability between devices of different bandwidths, such as 20 MHz and 40 MHz. For backward compatibility, a legacy header is placed in front of the new HT (High Throughput) packet format such that all the legacy devices will defer their transmission by decoding the legacy header. For 20/40-MHz interoperability, 40-MHz HT packets have both the legacy header and HT signal field duplicated in both 20-MHz channels so that 20-MHz devices in either 20-MHz channel can understand the header and defer their transmission accordingly. In addition, the 20-MHz HT packet format maintains exactly the same spectral footprint as the legacy 802.11abg systems. The design uses the same number of subcarriers as current 802.11abg systems, and hence does not require a new design for components to satisfy the spectral mask and adjacent channel rejection requirements. Also, the proposal dedicates the same 4 subcarriers as current systems to pilots for tracking the phase change, which implies the phase tracking performance will not be degraded with a similar phase noise specification in the current radio design. The TGn Sync proposal includes many examples of similar capabilities to reduce the stress placed by new designs on requirements such as the dynamic range of an ADC. In multiple antenna transmission, the power at the receiver can fluctuate significantly if the preamble design does not take ADC dynamic range into account. To address this, for example, the TGn Sync proposal uses a tone-interleaving technique to ensure that each section in the packet has a reasonably similar power at the receiver even for multipath channels.

New transmit diversity methods help the TGn Sync design combat the susceptibility of a wireless channel to multipath fading. This improves the robustness of the wireless link, even in the face of the additional demands of multiple antenna systems. These methods are especially important when AP is able to support more antennas and transmit more power than the clients. This reduces the demands and costs of client systems in the TGn Sync design. There are two distinct transmit diversity schemes in the TGn Sync proposal. One is orthogonal spreading with CDD (Cyclic Delay Diversity), which is effective for the case when the channel knowledge is not available to the transmitter. Since the same spreading is used for both training and data symbols (self-defining property), the receiver can decode data in the same manner as for the normal packet. For cases when the channel knowledge is available to the transmitter, the performance can be further improved by closed-loop transmit beamforming. The closed-loop beamforming scheme in the TGn Sync proposal is based on channel reciprocity, and most of the complexity increases can be allocated to the AP side with very little overhead to the clients. Furthermore, it also provides a simple packet-exchange mechanism for calibrating circuit mismatches between multiple chains if necessary.

Advanced coding techniques are also available in the TGn Sync proposal as optional enhancements to higher code rates using the existing convolutional code for higher data rates. These include LDPC (Low-Density Parity Check) codes and concatenated coding schemes with RS (Reed-Solomon) codes for improved robustness. By supporting different options for applications that need to optimize for either greater efficiency or greater reliability, this enhances the adaptability of 802.11n for non-traditional applications.

High precision guard intervals were motivated by actual field-measurements indicating that popular environments, such as the home, have small multipath rms delay spreads, which is often less than 50 ns. In such environments the 800 ns guard interval in 802.11a/g systems can be reduced to increase system efficiency and other measures of performance. The TGnSync PHY proposal introduces such a technique to increase data rates reliably for these environments.

This combination of MAC and PHY advances in the TGn Sync proposal dramatically improves the ability for 802.11 to scale to higher speeds, more applications, and the expansion of wireless spectrum. The core of 802.11 today is based on traditional computing applications, but it must also work well with the new demands for bandwidth placed by video, the demand on latency placed by voice, the demand on power placed by mobile, and with the new user capacity requirements coming from Wi-Fi's own popularity.