Adaptive interference canceller using analog algorithm with offset voltage
The changes in signal characteristics of many interference cancellation applications could be quite fast. They require the utilization of the adaptive algorithms. Algorithms such as Least Mean Square (LMS), Normalized Least Mean Square (NLMS) and Recursive Least Square (RLS) algorithms often have po...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://psasir.upm.edu.my/id/eprint/65479/1/FK%202015%20143IR.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The changes in signal characteristics of many interference cancellation applications could be quite fast. They require the utilization of the adaptive algorithms. Algorithms such as Least Mean Square (LMS), Normalized Least Mean Square (NLMS) and Recursive Least Square (RLS) algorithms often have poor numerical properties due to the practical implementation complexities. LMS and NLMS algorithms have been used in a wide range of signal processing applications because of their simplicity in computations compared to the RLS algorithm. The adaptation method of the LMS, NLMS and RLS algorithms quantizes the noise input signal during a limited period to track the error sample by sample. The output error signal is used for updating the weights processing the noise input signal. Since the quantization process is irreversible, an additive interference will be added to the input and output signals. This is to prevent losing the magnitude and sign information resulted from quantizing them within specific times. The output error signal used to update the weights causes an aliasing in the updated weights. This aliasing deviating the weights from their optimum values and increases the excess in the output Mean Square Error (MSE). This research describes a new approach for interference cancellation using an analog algorithm that adopts a new method to determine good tradeoff between the complexity and the convergence speed for optimum results. The proposed analog algorithm has been designed to deal directly with the continuous time domain of the input signal without quantizing it. Instead of using the error signal for updating the weights, the desired signal has been used to produce smooth weights to reduce the MSE output and to increase the convergence speed to the optimum output. In addition, the proposed algorithm has simple computational requirements that make it practically easy to be implemented. The experimental tests of the proposed algorithm have shown that the reduction percentage in the MSE output using the proposed algorithm is 51% compared to the NLMS algorithm and 61% compared to the RLS algorithm. In addition, the output Signal to Interference Ratio (SIR) using the proposed algorithm has been doubled compared to NLMS and RLS algorithms. |
|---|