Neural Memory Network for IoT Edge Data Management
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional memory systems are inefficient for IoT edge devices due to limited storage capacity, energy constraints, and bandwidth limitations, leading to challenges in processing high volumes of data with varying importance and complexity, especially when performing AI tasks.
Innovation Solution
An intelligent digital memory system utilizing a neural memory network with data neurons, cue neurons, gist neurons, spatial connections, and temporal connections, which dynamically reorganizes data based on access patterns and user requirements, and provides distributed spatio-temporal awareness for efficient data storage and retrieval.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Quantity of substance
If conventional memory systems are used in IoT edge devices, then device complexity is reduced, but storage capacity and data processing efficiency are insufficient
Solution Approach 1:
The memory system is segmented into multiple neural networks including spatio-temporal memory network, semantic memory network, and episodic memory network. Each network handles specific types of data processing and storage, allowing the system to manage large volumes of data while maintaining organized structure and efficient access patterns.
Solution Approach 2:
The patent implements nested memory structures where short-term memory is nested within working memory, which is nested within long-term memory. This hierarchical nesting allows efficient data management by storing frequently accessed data in faster memory layers while maintaining capacity for large-scale data retention in deeper layers.
2Quantity of substance
If high volume data is sensed and stored, then data completeness is improved, but energy consumption increases
Solution Approach 1:
The memory system assigns different quality levels and compression ratios to different data based on their importance and access frequency. Critical data is stored with high fidelity in accessible memory locations, while less critical data is compressed and stored in deeper memory layers, reducing overall energy consumption while maintaining data completeness.
Solution Approach 2:
The system performs preliminary data processing, filtering, and categorization at the edge device before storage. Data is pre-organized into semantic categories and temporal sequences, reducing the need for energy-intensive retrieval and processing operations later while maintaining complete and accurate data records.
3Loss of time
If data is stored with high retrieval speed, then access time is reduced, but storage capacity is limited
Solution Approach 1:
The hierarchical nested memory structure provides fast access to recently used data in short-term and working memory layers, while maintaining extensive storage capacity in long-term memory layers. The system automatically manages data transitions between layers based on access patterns, ensuring fast retrieval of critical data while preserving large-scale storage capability.
Solution Approach 2:
The memory system dynamically adjusts data placement and access paths based on temporal patterns and usage frequency. Frequently accessed data is automatically positioned in faster memory locations, while maintaining the ability to store and retrieve large volumes of data from deeper storage layers when needed.
4Quantity of substance
If redundant data is stored, then data completeness is improved, but storage efficiency decreases
Solution Approach 1:
The system creates selective copies of critical data across multiple memory locations and networks based on their importance and access patterns. Rather than storing all data redundantly, the system intelligently replicates only essential data elements, maintaining data completeness for critical information while improving storage efficiency by avoiding unnecessary duplication.
Solution Approach 2:
Different redundancy levels are applied to different data based on their semantic importance and access frequency. Critical data receives higher redundancy and replication, while less critical data is stored with minimal redundancy, optimizing the balance between data completeness and storage efficiency.
Data Source
AI summary
A computing entity comprising an intelligent digital memory system and one or more processors communicatively coupled to the intelligent digital memory system is provided. The one or more processors configured to receive one or more storage parameters, determine a store procedure cue neuron search location from candidate ones of a plurality of cue neurons associated with a neural memory network (NoK), insert the input data as a data neuron into the NoK based on the store procedure cue neuron search location, temporally link the data neuron with a location of last insertion, and modify the NoK in a manner of accessibility based on a pattern of a search for the store procedure cue neuron search location.