Abstract
In recent years, cognitive radios are regarded as a valid solution to solve the problem of inefficient spectrum utilization and increasing spectrum demand. A cognitive radio system which has the ability of spectrum sensing and transmission adjustment is proposed in this thesis. The spectrum sensing is an important issue because the spirit of cognitive radio is that secondary user’s transmission can not affect primary user. Therefore, an optimal detector applied for Thomson’s adaptive multitaper spectrum estimation (AMTSE) is presented. The detector is based on Neyman-Pearson Theorem and it can adjust the detection threshold according to the environment change. Also, this detector can reduce the number of observation time and has higher detection rate for a given false alarm rate compared with other methods. The simulation shows that the detection rate can outperform energy detection by 40%. Besides, it reduces the observation time about 60% than energy detection when achieving detection rate 0.9 at false alarm rate 0.001. Adjusting the transmission parameter is required after spectrum sensing. Here, a primary user which is 2×2 close-loop MIMO system and operates in TDD mode is taken into consideration. The assumption of TDD mode is that the secondary user can estimate the interference channel between its Tx and primary user’s Rx. By knowing the interference channel, block diagonalization which usually used in multiuser MIMO system is applied for canceling the interference when primary user exists. The limitation of block diagonalization is that the number of transmit antennas should be larger than the total number of receive antennas, so the secondary user needs extra 2 antennas for the cancelation of interference. By the precoding method, the interference to primary user can be eliminated when the CSI is perfectly known by secondary user. The capacity will exceed the Direct-Singular Value Decomposition (D-SVD) when SNR is larger than 14 dB and the SNR of primary user equals to 0 dB. If the primary user is idle, the secondary user’s Tx may can estimate the channel state information (CSI) due to the channel reciprocity. So the extra antennas can be used to antenna selection to improve the secondary user’s BER performance. Further, the AMTSE spectral detector is implemented by ASIC. We choose the DPSS number K = 2 to reduce the hardware cost. To deal with two paths of spectrum estimation, a novel FFT architecture is proposed. The 1024-point adopts the radix-2 and radix-2/4/8/16 to efficiently reduce the number of nontrivial twiddle factor multipliers. By using timing sharing techniques, the trivial twiddle factor multiplier can be realized by using some adders and shifters. The chip is implemented by TSMC 90 nm, and the power consumption of this detector is 30.9 mW at 122 MHz. The core area is 1057 × 1057 mm^2 and total chip area is 1557 × 1557 mm^2.