Research On Technologies Of Using M-sequences For Home Impulse Response Measurement

A maximal length sequence (m-sequence) is a pseudorandom binary sequence which has been widely exploited in areas of communication, radar and cryptography. During the last decades special attention also has been devoted to the study of technologies of using m-sequences to measure home acoustic impulse responses. There are two advantages in m-sequence technique, the one is its high noise immunity, and the other one is its quick computation speed and high efficiency. This dissertation has lucubrated many problems been met during the usage of this method. The key contributions of this dissertation are:1. The effects due to nonlinearity on m-sequence measurement are studied. We analyze the distortion in m-sequence measurement caused by nonlinearity with a Hammerstein structure. Because information of linear impulse response cannot be obtained by classical m-sequence method for an even order nonlinear Hammerstein system, an improved method has been presented. We propose an idea of using m-sequence with binary values 0 and 1 to stimulate a Hammerstein model system and acquired a fast improvement algorithm based on FMT of m-sequence with elements of 1 and -1 levels for measuring the linear impulse response. Home acoustic system with weak nonlinearity can be modeled by a simplified Volterra kernel model.According to this model, expressions of distorted linear impulse response corrupted by different order nonlinearity have been concluded. Factors which influence distortion immunity in m-sequence technique, for example, the period and amplitude of m-sequences and orders of nonlinearity, are investigated in detail, and many important results are also obtained.2. In order to improve distortion immunity, a method based on conventional truncation idea is proposed firstly, by which a truncation point can be determined accurately. Based on the relationship between second-order nonlinearity errors and third-order cross-correlation function of m-sequences, the truncation point can be simply obtained by choosing the peak value position with the smallest time lag of m-sequence's third-order cross-correlation function. The trinomial pairs of m-sequence with different original polynomial are different. To reduce the effect of nonlinearities selection of particular m-sequence is an important consideration. An idea is proposed that those m-sequences of which k 2 is big enough while k1 is smaller should be chosen so that those bigger 2-order nonlinearity impulses are far from linear impulse response and their effect on measurement can be decreased.3. As the length of Gold sequence is 2n-1, which is not directly suitable for FFT-based algorithm, a fast cross-correlation algorithm is expounded and improved.4. As few numbers in a preferred m-sequence maximal set cannot meet the need of simultaneous impulse response measurement, a kind of measurement exploiting balanced Gold sequences as test stimulus is proposed. Theoretic analysis and computer simulation results demonstrate the feasibility of the proposed method.