Practical Multi-access by Exploiting Spacial Diversity in 802.11bProject for ECE 734 2010 SpringHuimin Zeng1. INTRODUCTION The demand for wireless service is exponential increasing. Multi-access wireless LAN techniques, enabling users to share the resources, are one of the main components to alleviate this high demand. Four main multiple access technologies are used: frequency division multi-access, time division multi-access, code division multi-access, and space division multi-access. The first three have been extensively studied, while the spacial diversity has the risen attention because of emergence of the multiple-input and multiple-out (MIMO) technology. MIMO significantly boost network capacity by using multiple antennas at both the transmitter and receiver to exploit the spatial properties of wireless channels. MIMO has been incorporated into several wireless standards, including IEEE 802.11n. In single user MIMO system, the capacity improvement is bounded by the number of transmitting or receiving antenna, whichever comes smaller. In practice, the wireless devices generally have only a few antennas. For example, 802.11n devices only equip two antennas. However, access points (AP's) potentially can deploy a larger number of antennas. If we let multiple devices to form a “virtual MIMO” system, which is spacial multiple access in nature, then the network capacity would not be bounded the individual devices. Tse at. El [3] pointed out that the AP may be able to decode all concurrent frames correctly as long as the number of them is less than the number number of antennas at the AP. In this project, I am planning to design a practical system that enables multiple access by exploiting spacial diversity. Tan et. al. 's work, SAM [1], is my design baseline. 2. PROBLEM DESCRIPTIONOne of the main challenges in implementing spatial multiple access is the effect of intersymbol interference (ISI) when trying to decode concurrent transmissions from asynchronous senders in a wireless LAN. SAM applied interference alignment and cancellation [2] techniques to address this challenge. Let's look at a simple scenario, Illustration 1, where the AP is trying to decode two overlapped frames from two stations at different locations. The AP performs two steps to decode each frame: the interference nullifying (IN) and interference cancellation (IC).The idea of IN is to find two proper linear transforms for the two received signals, y1 and y2 in Illustration 2, at the two antennas such that, after the transformation, the components (x1 and x2) in these two received signals contributed by one transmitted frame are aligned with the the same direction and scale. Then a subtraction of these two transformed signals removes x1 and is ready x2 for decoding. In the next step, IC, the AP regenerates the original signal and cancels out the decoded x2, so that x1 can be decoded. In summary, the decode procedure is as following: (1) Use the clean preamble of the first arrived frame, Pa, of the two overlapped frames to align the signals.(2) Subtract the two signals to get the signal with the information of the second frame, Pb; decode Pb.Illustration 1: A simple scenarioST 1ST 2BSh11h22h21h12x1x2y1y2Illustration 2: Illustration of interference nullifyingy1h11* x1h21* x2h22* x2h11* x1y2y1h11* x1h21* x2R*h22* x2R*h11* x1R*y2(3) Re-encode Pb, and cancel it out from the original signal, to get the signal with the information of Pa; decode Pa. In this project, I am implementing IN and IC for 802.11b on a SDR platform. In the next session, I will describe the processing requirement for the implementation. 3. PROCESSING REQUIREMENT 3.1 802.11b PHY layer requirementThe PHY layer of 802.11b communication system contains four main functional blocks, as shown in Illustration 3. The binary bit stream at 2Mbps coming from the MAC layer will be first scrambled. This operation does not change the bit rate. The bit stream then is fed to QPSK modulation, where every two bits are mapped to one complex symbol. To provide sufficient fidelity, these complex samples require 16-bit quantization for both I and Q components, therefore 32 bits per symbol and the bit rate coming out of QPSK is at 32Mbps. Next operation is Direct Sequence Spread Spectrum (DS-SS) to allow 802.11b provide different data rate. The code length used in 802.11b is 11 chips, so the bit rate coming out of DS-SS will be increased by 11 times. The last operation is up sampling by a factor of four. This is a technique widely used for better performance [6]. Therefore, finally we have 1.4Gbps out of the PHY layer. Illustration 3: PHY layer in 802.11b transceiverScrambleQPSKModDS-SSUp sampling2Mbps 32Mbps 352Mbps1.4Gbps2MbpsMACTransmissionDe-ScrambleQPSKDemodDS-SSDecodeDown sampling2Mbps 32Mbps 352Mbps1.4Gbps2MbpsMACReceptionDecode PbRegenerate PbDecode PaAs for the reception, in order to enable spacial multi-access as described above, even just for the simplest scenario – decoding two overlapped frames, it will require to go the those four basic functional blocks three times: decoding Pb, regenerating Pb and finally decoding Pa. We can see that the processing requirement for implementing the spacial multi-access is high. Firstly, the interface must be able to sustain 1.4 Gbps throughput. Conventional gigabit ethernet card cannot meet this requirement. Secondly, the computation intensity is high. If there are N operations per bit, this will require 1.4 times N Giga operations per second. Finally, real-time communication requires real-time processing – the processing latency needs to meet response deadlines. I choose Sora platform to implement this project. Its radio control board has a maximum throughput of PCIe X32, which is equivalent to 64 Gbps. Its software also support both SIMD instructions and multi-core processing. 3.2 Sora PlatformSora is a fully programmable software defined radio (SDR) on commercial personal computer architectures. Sora platform is composed of both software and hardware. The hardware components consist of a radio front-end for reception and transmission and a radio control board for high-throughput data transfer between the radio and the host memory storage. Sora offers dedicated CPU cores programming and SIMD processor extensions for high speed PHY layer processing on general purpose processor. Detailed information about Sora and an example project, SoftWiFi, can be found in [4].
View Full Document
Unlocking...