Systolic RLM Array Reduces Neural Signal Data Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional implanted neural devices face challenges in processing and storing large volumes of neural signal data from multiple sensors, often exceeding the processing bandwidth of controllers, leading to power, area, and throughput issues, with redundant or irrelevant data present in the signals.
Innovation Solution
A low-power hardware architecture using a systolic random-logic-macro (RLM) array combines serially sampled neural signal data with a transformation matrix (TFM) to select, combine, or suppress signal data, generating a reduced data set that represents relevant neural activity characteristics, and dynamically adjusts based on changes in neural activity and system parameters.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If the number of sensors is increased to improve neural signal measurement coverage, then measurement precision and neural activity characterization are improved, but data volume and processing complexity increase beyond controller bandwidth
Solution Approach 1:
The patent segments the data processing function by introducing an intermediate processing stage between sensor acquisition and controller analysis. The neural signal data is divided into multiple data channels that are processed in parallel through the RLM array, with each RLM handling specific transformation operations. This segmentation allows the system to manage large sensor arrays without overwhelming the controller.
Solution Approach 2:
The patent introduces an intermediary processing architecture consisting of the RLM array and transformation matrix that sits between the sensor array and the controller. This intermediary performs real-time linear transformations on the neural signal data, reducing dimensionality and extracting relevant features before data reaches the controller, thereby bridging the gap between high-volume sensor output and limited controller bandwidth.
2Loss of information
If all sensor data is processed to ensure complete neural activity characterization, then data completeness is improved, but power consumption and processing time increase
Solution Approach 1:
The patent extracts only the most relevant neural signal information by applying linear transformations through the RLM array that identify and preserve significant signal components while discarding redundant or noisy data channels. The transformation matrix is configured to extract essential neural activity characteristics, allowing the system to maintain information completeness for meaningful analysis while eliminating unnecessary data that would consume processing power.
Solution Approach 2:
The patent applies partial processing by selectively activating and configuring RLM elements based on the specific neural recording needs. Not all possible data transformations are performed - only those necessary for the current experimental or clinical objective. This partial action approach maintains sufficient information completeness for the specific application while avoiding the excessive power consumption that would result from processing all possible data combinations.
3Device complexity
If a fixed processing architecture is used to simplify system design, then device complexity is reduced, but adaptability to different neural recording configurations is limited
Solution Approach 1:
The patent implements a dynamic processing architecture where the RLM array can be reconfigured through programmable transformation matrices. The system allows dynamic adjustment of the number of active RLM elements, the dimensions of the transformation matrix, and the specific linear transformation operations performed. This dynamic capability enables the same hardware architecture to adapt to different sensor counts, sampling rates, and neural recording configurations without requiring physical reconfiguration or redesign.
Data Source
AI summary
Described herein are systems and methods for reducing the size of data payloads delivered to downstream processing from a raw series of biological sensor recordings. In one variation, the system comprises a low-power hardware architecture that combines serially sampled neural signal data with a transformation matrix (TFM) using a novel systolic random-logic-macro (RLM) array.