Vehicle-Shared Memory Pools for Data Center Bandwidth
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in memory bandwidth constraints due to limitations in I/O technology, power, and packaging, which can impact safety and performance, and existing solutions often compromise reliability for reduced component count.
Innovation Solution
A data center using a memory pool between selected memory resources, allowing vehicles to wirelessly access and share memory resources from a pool formed by multiple vehicles or base stations, optimizing memory usage and reducing reliance on individual vehicle components.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If component count is reduced to decrease system failure risk, then reliability improves, but performance requirements cannot be met
Solution Approach 1:
The patent combines memory resources from multiple autonomous vehicles into a shared memory pool, allowing vehicles to access pooled memory capacity rather than relying solely on individual vehicle memory. This merging approach increases total available memory and bandwidth without requiring each vehicle to carry redundant memory components, thus maintaining performance while potentially improving reliability through resource sharing.
Solution Approach 2:
The memory pool serves multiple functions: it provides memory storage, enables data sharing between vehicles, and offers bandwidth resources to multiple vehicles simultaneously. This multi-functional approach allows a single pooled resource to replace what would traditionally require multiple separate memory systems across different vehicles.
2Productivity
If memory bandwidth is increased to meet AI/ML requirements, then performance improves, but power consumption and thermal management become problematic
Solution Approach 1:
Multiple vehicles contribute their memory resources to a shared pool, collectively providing high memory bandwidth that would be impossible for a single vehicle to achieve with its own components. This distributed approach spreads the power and thermal load across multiple vehicles rather than concentrating it in one system.
Solution Approach 2:
Each vehicle in the pool contributes its own memory resources to the shared pool, effectively serving itself and others simultaneously. The memory resources are already powered and managed by individual vehicle systems, eliminating the need for additional dedicated power infrastructure for the memory subsystem.
3Productivity
If more memory resources are added to individual vehicles, then performance improves, but device complexity and packaging constraints are violated
Solution Approach 1:
The patent transitions from a single-vehicle memory architecture to a multi-vehicle distributed memory architecture, adding the dimension of spatial distribution across multiple vehicles. This allows the system to achieve high memory capacity without increasing the complexity or packaging requirements of any single vehicle's memory subsystem.
Data Source
AI summary
Apparatuses, systems, and methods related to a data center using a memory pool between selected memory resources are described. A data center using a memory pool between selected memory resources may enable performance of functions, including automated functions critical for prevention of damage to product, personal safety, and/or reliable operation, based on increased access to data that may improve performance of a mission profile. For instance, a method described herein includes transmitting, from a processor at a first vehicle that comprises the processor and memory, a request to access a pool of memory resources configured from a plurality of vehicles each having a local processor and memory, receiving, from a second vehicle of the plurality of vehicles, an indication to access the pool of memory resources, and reading data from or writing data to the memory at the second vehicle using the processor at the first vehicle based at least in part on receiving the indication to access the pool of memory resources.