Physical Layer Approach for 802.11n doc.: IEEE 802.11-04/0abcr0 July 2004 Physical Layer Approach for 802.11n Mustafa Eroz, Feng-Wen Sun, Lin-Nan Lee & Mallik Moturi Hughes Network Systems 11717 Exploration Lane Germantown, MD 20876 Mustafa Eroz, Hughes Network Systems Mustafa Eroz, Hughes Network Systems

802.11n Physical Layer Design Issues July 2004 802.11n Physical Layer Design Issues Achieving more than 100 Mbps information throughput in 20MHz channel without using additional power/bandwidth or accepting range compromise is very difficult This is a 802.11n mandatory requirement 802.11a/g sends 54 Mbps at “top speed”, information throughput is about 60% of that. Assuming similar MAC and Link Layer overhead, we need to be about 3-4 times more efficient than 802.11 a/g. This translates to about 11-12 bits per channel use with channel coding and other overhead, comparable to sending 56 kbps information through a telephone line. In the wireless channel, this kind of transmission efficiency can only be achieved with some form of Multi-Input/Multi-Output (MIMO) space-time diversity scheme with more than two transmit antenna. Mustafa Eroz, Hughes Network Systems

Technical Challenges of MIMO Transmission July 2004 Technical Challenges of MIMO Transmission To meet safety and interference regulations, total transmit power must be the same as single transmit antenna case Early MIMO schemes such as Blast, Space-Time codes, and their variations require very high SNR to achieve acceptable packet error rate performance, leading to very short range for required bandwidth efficiency The required information throughput at reasonable transmit power can only be achieved in conjunction with very powerful error correction coding schemes Near Shannon-limit codes have been standardized by the 3rd Generation Wireless (3GPP & 3GPP2) for mobile wireless application in 1999, and Digital Video Broadcast (DVB) project for direct broadcast via satellite recently (2003) We expect near Shannon-limit codes redesigned for the MIMO architecture and adapted to the general 802.11 framework to provide the only solution which meets this challenge. Mustafa Eroz, Hughes Network Systems

Forward Error Correction (FEC) Coding as the Enabling Technology doc.: IEEE 802.11-04/0abcr0 July 2004 Forward Error Correction (FEC) Coding as the Enabling Technology For the past several decades, forward error correction coding has been responsible for the most significant improvements for many communication systems. Mustafa Eroz, Hughes Network Systems Mustafa Eroz, Hughes Network Systems

July 2004 Applications of State-of-the-Art FEC Codes to Consumer Marketplace – Milestone Examples Design of turbo coding techniques for mobile wireless channels and adoption of optimized codes, interleaver, and puncturing patterns for 3GPP and 3GPP2 standards (1999). Major breakthrough since Viterbi decoding (1968) These codes are implemented in more than 50 million cdma2000 and W-CDMA/UMTS cellular phones already Adoption of high-performance low-density parity check (LDPC) codes for DVB-S2 standard (2003) Shannon capacity for SISO channel has been reached for all practical purposes. These codes have been implemented in silicon and will be deployed in next generation broadcast and interactive satellite networks shortly. Mustafa Eroz, Hughes Network Systems

Turbo Codes for 3G Wireless July 2004 Turbo Codes for 3G Wireless Turbo codes provides substantial gain in third generation wireless systems (v=30 kph) Mustafa Eroz, Hughes Network Systems

July 2004 3G Turbo Code Design Even though the turbo codes were known a few years earlier, key aspects must be designed Constraint length and best generator polynomials for the constituent code Best puncturing tables for all the required code rates A flexible algorithmic turbo interleaving technique with guaranteed distance criteria for any interleaver sizes, avoiding separate interleaver table for each block size All these aspects were optimized, specified, and adopted by 3GPP and 3GPP2 Mustafa Eroz, Hughes Network Systems

July 2004 LDPC Codes Low density parity check (LDPC) codes offer a much richer platform for various implementation techniques than turbo codes. While any randomly chosen turbo code would have reasonably good performance, large block size LDPC codes have a greater potential to approach Shannon limit. But, a great deal of expertise is needed to design such LDPC codes. Similarly, if the parity check matrix design of LDPC codes is done with no restrictions, the resulting decoder complexity would be high. DVB-S2 LDPC codes provide performance close to Shannon limit while avoiding decoder complexity. Mustafa Eroz, Hughes Network Systems

LDPC Advantages Excellent performance July 2004 LDPC Advantages Excellent performance Very simple LDPC decoder implementation More suitable to high speed applications since LDPC decoder operations can be performed in parallel. Adoptable to any higher-order modulation with simple mapping. Mustafa Eroz, Hughes Network Systems

July 2004 What is LDPC Code ? LDPC codes are linear block codes with sparse parity check matrix H(n-k)xn An example LDPC code with n=8 and R=1/2 Bit nodes and check nodes communicate with each other iteratively to find the transmitted values of bit nodes. Mustafa Eroz, Hughes Network Systems

Simple LDPC Encoding Restrict the parity check matrix to where July 2004 Simple LDPC Encoding Restrict the parity check matrix to where Encode information block into codeword Using and recursively solving for parity bits : Solve Mustafa Eroz, Hughes Network Systems

LDPC Decoder Algorithm July 2004 LDPC Decoder Algorithm Demod output: where N is the block size ( log-likelihood ratio of received bits) Initialization: Mustafa Eroz, Hughes Network Systems

Check Node Update It can be shown that July 2004 Mustafa Eroz, Hughes Network Systems

Bit Node Update & Hard Decision July 2004 Bit Node Update & Hard Decision Bit Node Update: Hard Decision: Stop the iterations if hard decisions satisfy all the parity check equations or if the maximum number of iterations is reached. Mustafa Eroz, Hughes Network Systems

Performance in AWGN Channels July 2004 Performance in AWGN Channels Mustafa Eroz, Hughes Network Systems

Up to 40% Throughput Improvement over DVB-S1 July 2004 Up to 40% Throughput Improvement over DVB-S1 Mustafa Eroz, Hughes Network Systems

LDPC Codes for MIMO Channels July 2004 LDPC Codes for MIMO Channels Advanced LDPC codes bring the performance of practical communication system very close to theoretical limits for single-input, single-output AWGN. With clever customization and optimization, LDPC codes can approach Shannon limit for MIMO fading channels as well. We intend to submit a physical layer proposal based on a set of LDPC codes highly optimized for 802.11n application before the next meeting. Mustafa Eroz, Hughes Network Systems