Tone Detection Using Convolution Impulse Responses
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing tone detection methods in digital signal processing, such as DFT and Goertzel methods, are inefficient due to high clock cycle requirements, limiting processor utilization and concurrent channel processing capabilities in digital signal processors (DSPs) due to memory bus bandwidth constraints.
Innovation Solution
The novel tone detection method employs convolution techniques using pre-computed impulse responses to calculate tone power, reducing the number of clock cycles needed by calculating y1(N−1) and y1(N−2) through summation of samples multiplied by filter coefficients, rather than recursive calculations, thereby optimizing processing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If DFT or Goertzel methods are used for tone detection, then tone detection functionality is achieved, but the number of clock cycles required is high (15 or over 40 clock cycles)
Solution Approach 1:
The patent segments the tone detection process into two separate convolution operations (first value calculation and second value calculation) that can be executed in parallel. Each convolution operation processes a different aspect of the signal (real and imaginary components), allowing simultaneous computation that reduces total execution time from over 40 clock cycles to approximately N/4 clock cycles.
Solution Approach 2:
The patent transforms the recursive Goertzel algorithm into a convolution-based approach, changing the computational dimension from sequential recursion to parallel convolution operations. This dimensional change enables the use of pre-computed impulse responses and allows multiple calculations to proceed simultaneously, dramatically reducing the clock cycles required while maintaining detection accuracy.
2Measurement precision
If DFT or Goertzel methods are used for tone detection, then tone detection functionality is achieved, but processor utilization is limited due to memory bus bandwidth constraints
Solution Approach 1:
The patent pre-computes and stores the impulse response coefficients before the actual tone detection process. By having these coefficients readily available in memory, the processor avoids repeated calculations and can directly apply them during convolution operations, significantly improving processor utilization and reducing memory bus bandwidth constraints during the critical detection phase.
Solution Approach 2:
The patent enables continuous processor utilization by structuring the tone detection as efficient convolution operations that can be pipelined and executed without idle cycles. The parallel computation of first and second values keeps the processor continuously engaged, maximizing productivity while maintaining accurate tone detection.
3Measurement precision
If DFT or Goertzel methods are used for tone detection, then tone detection functionality is achieved, but the number of concurrent channels that can be processed is limited
Solution Approach 1:
The patent segments the computation into independent convolution operations that can be parallelized across multiple channels. Each channel's tone detection can proceed simultaneously using the same pre-computed impulse responses, enabling the processor to handle many more concurrent channels without sacrificing detection accuracy.
Solution Approach 2:
The patent creates a universal tone detection framework where the same convolution-based algorithm and pre-computed impulse responses can be applied to multiple channels simultaneously. This multi-functional approach allows the processor to efficiently handle various telephony applications (DTMF detection, echo canceller control, dial tone detection, fax negotiation) across numerous concurrent channels.
Data Source
AI summary
An apparatus for determining the presence of a tone in an input signal includes memory circuitry and data processing circuitry coupled to the memory circuitry. The data processing circuitry is operative to receive multiple samples of the input signal, and to determine a first value at least in part by multiplying each of the samples by respective ones of a first set of values for an impulse response and summing the results. The data processing system is also operative to determine a second value at least in part by multiplying each of a portion of the samples by respective ones of a second set of values for the impulse response and summing the results. The data processing system is operative to determine the power of the tone in the multiple samples of the input signal at least in part by utilizing the first value and the second value.