LMS MFU Algorithm for Acoustic Imaging

We consider the problem of secondary source feedback in acoustic imaging with a reference microphone. While the LMS MFU algorithm has a reasonable convergence performance, the newly proposed MFU-LMS algorithm shows an improved convergence performance. The model results include the estimated impulse response and updated filter coefficients.

Adaptive LMS algorithm

In this paper, an Adaptive LMS algorithm for mFu is present. It is an improvement on the conventional LMS algorithm as it provides the same output and error signals. It also includes an estimated secondary path SdzTh. The results of the algorithm were compare with other algorithms. The algorithm parameters are list in Table 2. The best performing parameters were select by trial and error.

An adaptive LMS algorithm has the ability to reduce mean-square error and convergence coefficient. This is possible by optimizing filter weights. An optimal weight is the one that minimizes the mean-square error, and a LMS algorithm approaches it by ascending and descending the mean-square-error vs. filter weight curve.

The MFU-LMS algorithm incorporates an adaptive hyper-stable recursive filter with variable step size. It has been applied to two experimental cases: a sinusoidal signal buried in white noise, and a chirp signal embedded in white noise. The results show that the algorithm is capable of fast convergence and stability, and it improves sensitivity in the chirp signal.

Variable step size

A variable step size LMSU algorithm is an improvement over the fixed-step LMS adaptive filter algorithm. It eliminates irrelevant noise and improves anti-interference capabilities of the adaptive filter. It also has the advantage of providing more flexibility for practical applications. This paper discusses the merits of the variable step size LMSU algorithm.

The first step is to select an appropriate VSS algorithm, which controls the step size. A variety of VSS algorithms can used to control the step size, as long as they are noise-constrained. We present two variations of this algorithm: the noise-constrained LMS algorithm (NC), and the variable step-size quotient LMS algorithm (VSQ).

In addition to reducing the computational requirements, the variable step-size algorithm can also reduce the error rate and the convergence rate of an LMS algorithm. We compare the performance of VS-LMS algorithms that use a variable step-size algorithm and their analytical results.

Comparison with other conventional algorithms

Traditional algorithms require well-defined inputs and outputs, can be implemented in a finite number of steps, and must find a solution with available resources. Moreover,Click they must complete a task after a finite number of steps. For example, a traditional algorithm might solve a large problem by breaking it into smaller subproblems and solving each of them independently. This method is commonly call the Divide and Conquer algorithm.

Application to a short duct

The correlation method has been the traditional method for impact location detection. This method uses the time delay between acoustic waves propagating in a duct. This information is use to estimate the impact location. It is important to note that the acoustic wave should propagate with nondispersive characteristics.

Show More

Related Articles

Back to top button