Time-Encoded Signal Processing With Spike-Pulse Feedback Loops
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing signal processing technologies face limitations in accuracy due to dynamic range constraints in analog computing and introduce quantization noise in digital domains, particularly when dealing with real-time processing of analog signals from RF or hyperspectral sensors.
Innovation Solution
A nonlinear network that combines spike domain and pulse domain processors, utilizing 1-bit or 1.5-bit Digital to Analog Converters (DACs) for internal feedback loops, allowing for time-encoded signal processing without quantization noise, and incorporating converters to interconnect spike and pulse domain signals, enabling efficient real-time processing.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If analog amplifiers are used for signal processing, then processing speed is fast, but accuracy is severely limited by dynamic range of the analog components
Solution Approach 1:
The patent introduces time encoding as an intermediary mechanism that converts analog signals into spike trains where information is represented by timing rather than amplitude. This mediator allows the system to avoid the dynamic range limitations of analog amplifiers while maintaining processing speed, as the time-encoded signals can be processed by simple comparators and logic gates without requiring high-precision analog components.
Solution Approach 2:
The patent replaces the mechanical/analog amplifier system with a digital-like time-encoded processing system. Instead of using analog amplifiers with limited dynamic range, the system uses time-encoded spike signals that can be processed by simple electronic components such as comparators, integrators, and logic gates, thereby substituting the analog amplification mechanism with a time-based processing mechanism that has superior precision.
2Measurement precision
If ADC conversion is used to digitize signals, then processing accuracy improves, but speed is limited by the performance of ADC conversion and quantization noise is introduced
Solution Approach 1:
The patent performs time encoding as a preliminary action before the signal enters the processing network. By converting the analog signal into a time-encoded spike train at the input stage, the system avoids the need for subsequent ADC conversion during processing. This preliminary time encoding enables the signal to be processed in the time domain without the speed limitations and quantization noise associated with traditional ADC-based digitization.
Solution Approach 2:
The patent changes the representation parameter of the signal from amplitude-based (analog) or binary-digit-based (ADC output) to time-based (spike timing). This parameter change allows the signal to carry analog information in the timing of spikes rather than in amplitude levels or digital codes, thereby achieving both high precision and fast processing without the trade-offs of traditional ADC conversion.
3Measurement precision
If traditional digital circuits with timing gates are used, then quantization noise is reduced, but timing transitions are quantized by the presence of timing gates
Solution Approach 1:
The patent uses dynamic, asynchronous timing gates that adapt to the incoming spike rates rather than operating at fixed frequencies. The timing gates are designed to be responsive to the actual signal dynamics, allowing them to pass timing information with minimal quantization. This dynamic approach contrasts with static timing gates that impose fixed temporal constraints and cause information loss.
Solution Approach 2:
The patent incorporates feedback mechanisms in the processing cells where the output spikes are fed back to influence the timing of subsequent processing. This feedback allows the system to maintain precise timing information by continuously adjusting the processing based on the actual spike timing, thereby reducing the quantization effects that would otherwise be introduced by fixed timing gates.
Data Source
AI summary
A method of, and apparatus for, the processing analog data. The method includes the steps of: time encoding the analog data; setting weighting factors in an array of 1-bit DACs for processing the time encoded analog data; summing and filtering outputs of the array of 1-bit DACs; time encoding the filtered outputs of the outputs of the array of 1-bit DACs; and coupling the time encoded filtered outputs and analog input data into inputs of the array of 1-bit DACs.