Data Storage Controller ML Models with Certificate-Based Updates

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing data storage devices lack efficient mechanisms for storing and updating machine learning models tailored to individual customer needs, leading to suboptimal performance due to one-size-fits-all ML models.

Innovation Solution

Implementing a data storage device with a memory configured to store multiple ML models and customer digital certificates, allowing customers to access, update, and replace models securely using secure enclave technology and customer digital certificates, optimizing data relocation and prefetching.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If a data storage device stores multiple ML models and enables customer access via digital certificates, then customer-specific performance optimization is improved, but device complexity increases

Engineering Contradiction:

Improvecustomer-specific model customizationVSAvoidsecurity framework and model management

Core Design Contradiction:

Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent segments ML models into distinct entries within the storage device memory, each associated with specific customer digital certificates. This allows the system to store multiple customer-specific models separately and selectively retrieve only the relevant model based on the presented certificate, thereby enabling customization without overwhelming complexity in model management

Inventive Principle:

Principle #1Segmentation

Solution Approach 2:

The patent introduces digital certificates as intermediary elements that mediate between customers and ML models. These certificates serve as secure identifiers that the storage device uses to authenticate customers and determine which model should be retrieved and executed, simplifying the access control mechanism while enabling personalized service

Inventive Principle:

Principle #24Intermediary (Mediator)

2Productivity

If the storage device retrieves and loads ML models into processing components, then model execution performance is improved, but data access time increases

Engineering Contradiction:

Improvedata processing speedVSAvoidmodel retrieval and loading time

Core Design Contradiction:

ProductivityVSLoss of time

Solution Approach 1:

The patent implements preliminary action by pre-storing multiple ML models in the storage device memory before they are needed. When a customer presents their digital certificate, the system can immediately retrieve the pre-loaded model without requiring time-consuming model generation or training at the moment of execution, thus reducing access time while maintaining high processing speed

Inventive Principle:

Principle #10Preliminary action

Data Source

PatentUS20250298880A1Storage methods and devices with secure customer-updatable machine learning models

Publication Date: 2025.09.25 SANDISK TECHNOLOGIES LLC

AI summary

A data storage device stores a set of pre-trained machine learning (ML) models along with corresponding customer digital certificates and other information. Techniques are provided herein for allowing customers to access selected ML models using a digital certificate and a public key so that the selected ML models may then be loaded into a data storage controller of the data storage device for use, for example, by firmware of the controller to intelligently control data pre-fetch, data relocation, or other data processing functions. Techniques are also provided to allow customers to selectively update or replace ML models with customer-supplied models, subject to authentication of the customer and verification of the compatibility of the ML model within the data storage controller. Still other techniques are provided for allowing ML models from the data storage device to be loaded into a secure enclave within the host to execute within the secure enclave.