Design of Cost Efficient FEQ for IEEE802.16a OFDM System

在這個論文中，提出一些應用於IEEE802.16a 正交分頻多工調變系統的設計，像是低功率的頻域等化器，利用DC次載波偵測去找到整數倍載波飄移和通道內插的分析。 低功率頻域等化器是根據strength-reduced 複數乘法器的觀念去設計，讓整個硬體上的需求降低，提出的低功率等化器比傳統的架構節省了將近19%的運算量。由於在作通道估計時需要用到ROM去儲存已知的前置碼，為了降低硬體成本，一個藉由控制MUX 來取代大量ROM 的方法被應用在系統。除此之外，選擇single port RAM 代替dual ports RAM來存取在LMS adaptation會被讀寫的訊號，並且利用合併的技巧，將兩個獨立的RAM合成一塊，使控制訊號只要一組就足夠。使用以上技巧，在ROM方面：面積可以省70%，而在功率消耗量上可節省53%。就RAM 而言，面積省20%，功率消耗量節省18.4%。 整數倍載波飄移的設計，用DC次載波找飄移量比傳統用matched-filter 來找，就硬體成本而言節省了88.6%，功率消耗量方面省了85%。 在802.16a 正交分頻多工調變系統，由於通道估計只能在偶數次載波上通道資訊，在奇數次載波就必須利用已知的通道去作內插估計。越複雜的內插方法雖然會有比較好的精準度，但所需實現的硬體需求也相對較高。在此論文中，針對MSE 對Linear 與Piecewise-Parabolic 內插提出一套完整的分析，根據分析、模擬結果與系統的考量，最後選擇Linear 內插來作奇數次載波上通道的估測。最後有對低功率頻域等化器作c code 浮點數symbol error rate 的模擬，由分析可看出這樣的設計符合系統的要求。In this thesis, the design of IEEE 802.16a OFDM-based WiMAX systems is presented. The system design includes the cost efficient FEQ, dc-tone power detection integer CFO estimation, channel interpolation analysis and the CFO tracking loop. According to strength-reduced complex-value multiplier, the cost efficient FEQ is proposed to decrease the hardware cost. The hardware cost of proposed FEQ is reduced 19% compared with the conventional FEQ. On the other hand, in order to reduce the ROM size in the channel estimation to store the preamble data, the control-MUX method is proposed. In addition, considering the storage of the channel coefficient and decision error in the LMS adaptation, the single port RAM is selected to decrease the RAM size requirements. Furthermore, combing the two RAM blocks, channel coefficient RAM and decision error RAM, to a single RAM is applied to the cost efficient FEQ design. Consequently, the total size for the ROM and RAM can be reduced by 70% and 20% respectively. Besides, the power consumption is also saved by 53% and 18.4% individually. Considering integer CFO estimation, the dc-tone power detection in the frequency-domain is proposed to achieve the low-computing estimation. Therefore, the approach can save 85% power and 88.6% hardware cost compared with the matched-filter based integer CFO estimation. The channel interpolation is necessary in WiMAX systems to obtain the channel information on the odd tones since the preamble only provide the even tones information. Although the complicated interpolation method, Raised-Cosine interpolation, can minimize the estimation error, the method results in the higher hardware cost simultaneously. Therefore, in order to acquire the best trade-off between performance and hardware cost, the estimation error of the Linear and Piecewise-Parabolic channel is analyzed based on the MSE criterion. According to the analysis results, the Linear and Piecewise-Parabolic interpolation satisfy to system requirement. From the hardware cost point of view, the Linear channel interpolation is adopted for the cost efficient FEQ. Finally, the floating and fixed point SER simulations are presented to meet the system requirements.1 Introduction 1 1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 1.2 Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 1.3 Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 2 Wireless MAN System and Background Knowledge 7 2.1 Wireless MAN Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 2.2 OFDM Principles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 2.2.1 OFDM Signal Modeling . . . . . . . . . . . . . . . . . . . . . . . 9 2.2.2 Guard Interval and Cyclic Pre . . . . . . . . . . . . . . . . . . 13 2.3 IEEE 802.16 Standard and Specications for OFDM Mode . . . . . . . . 13 2.3.1 Main Parameters . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 2.3.2 Preamble Format . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2.3.3 Structure of the down-link frame . . . . . . . . . . . . . . . . . . 17 2.3.4 Adaptive Modulation and Forward Error Correction . . . . . . . . 18 2.3.5 Pilot Modulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 3 OFDM System and Wireless Channel Model 21 3.1 OFDM System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 3.2 Wireless Channel Model . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 3.2.1 Path Loss(PL) and Shadowing . . . . . . . . . . . . . . . . . . . . 23 3.2.2 Multipath Fading . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 3.2.3 Power Amplier Nonlinearity . . . . . . . . . . . . . . . . . . . . 25 3.2.4 Power Delay Prole (PDP) . . . . . . . . . . . . . . . . . . . . . . 25 3.2.5 Time Dispersion . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2.6 Coherence Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.7 Doppler Shift . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 3.2.8 Coherence Time . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 3.2.9 Rayleigh Fading and Ricean Fading . . . . . . . . . . . . . . . . . 30 3.2.10 Channel Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 4 Synchronization Techniques in OFDM Systems 33 4.1 Symbol Boundary O set . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.1.1 Delay Correlator (Coarse Symbol Boundary Detection) . . . . . . 35 4.1.2 Matched Filter (Fine Symbol Boundary Detection) . . . . . . . . 36 4.2 Carrier Frequency O set . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 4.2.1 Fractional CFO Estimation (Correlator) . . . . . . . . . . . . . . 38 4.2.2 Integer CFO Estimation . . . . . . . . . . . . . . . . . . . . . . . 39 4.2.3 CFO Tracking Loop . . . . . . . . . . . . . . . . . . . . . . . . . 45 5 Equalization 51 5.1 Channel E ects in Time Domain and Frequency Domain . . . . . . . . . 51 5.2 Equalization of the OFDM system . . . . . . . . . . . . . . . . . . . . . . 52 5.2.1 Channel Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 53 5.2.2 Interpolation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54 5.2.3 Interpolation Analysis . . . . . . . . . . . . . . . . . . . . . . . . 56 5.2.4 Adaptive . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67 6 FEQ Design 71 6.1 Conventional FEQ and Cost E cient FEQ . . . . . . . . . . . . . . . . . 72 6.1.1 Filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 6.1.2 Updating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 6.1.3 Channel Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 75 6.1.4 Comparisons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 6.2 Hardware Design for Cost E cient FEQ . . . . . . . . . . . . . . . . . . 80 6.2.1 Simulation for Signal Word length . . . . . . . . . . . . . . . . . . 80 6.2.2 Channel Estimation . . . . . . . . . . . . . . . . . . . . . . . . . . 84 6.2.3 Slicer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86 6.2.4 Updating . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90 6.3 FPGA Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 7 MISO 99 7.1 Transmitter Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 7.2 Channel Estimation for MISO system . . . . . . . . . . . . . . . . . . . . 100 8 Conclusion 103 Bibliography 10