Data Processing Circuit Network Arbitration via Message Seniority
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In real-time data processing circuits, ensuring guaranteed communication bandwidth for streams without storage of reservation information in node circuits is challenging, leading to potential interruptions and increased circuit area, especially when interconnecting growing numbers of data processing units.
Innovation Solution
A time-slot multiplexing network with node circuits that arbitrate message access based on seniority, ensuring that once an initial message is forwarded, subsequent messages are prioritized, reducing the need for storage and minimizing resource contention by using confirmations to establish guaranteed access.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If centralized planning or reservation programming is used to guarantee bandwidth, then reliable real-time communication is achieved, but node circuit complexity and area increase due to storage requirements
Solution Approach 1:
The patent extracts the reservation information from the node circuits and places it in the message packets themselves. Each message carries its own reservation data, allowing nodes to make forwarding decisions without storing reservation state. This resolves the contradiction by maintaining reliability through reservation guarantees while eliminating the complexity and area overhead of storage in node circuits.
Solution Approach 2:
Messages carry their own reservation information and path identification, enabling them to self-identify and secure bandwidth guarantees without external booking. The message itself contains all necessary information for nodes to provide guaranteed service, eliminating the need for centralized planning or node-based reservation storage.
2Reliability
If buffer memory is added to nodes to mitigate contention, then stream continuity is improved, but circuit area increases
Solution Approach 1:
The patent removes the need for buffer memory in nodes by extracting the contention resolution mechanism from the node infrastructure and embedding it in the message packets. Reservation information in the messages allows nodes to forward packets without buffering, maintaining stream continuity while avoiding the area overhead of buffer memory.
Solution Approach 2:
Reservation information is established in advance and embedded in each message before transmission. This preliminary action allows nodes to immediately forward messages without needing to buffer them for contention resolution, as the reservation guarantee is already encoded in the message itself.
3Reliability
If more messages are transmitted to establish stream reservation, then bandwidth guarantee reliability is improved, but communication setup time increases
Solution Approach 1:
The first message of a stream carries complete reservation information that pre-establishes the bandwidth guarantee for all subsequent messages. This preliminary action in the first message eliminates the need for multiple setup messages, reducing stream setup time while maintaining reliability through the embedded reservation data.
Solution Approach 2:
The patent changes the parameter of reservation information from being stored in node memory to being transmitted as data in message packets. This parameter change allows reservation guarantees to be established instantly in the first message without requiring multiple exchange messages, reducing setup time while maintaining the reliability of bandwidth guarantees.
Data Source
AI summary
A data processing circuit on an integrated circuit and an associated method of processing data in the data processing circuit is provided, the data processing circuit comprising a network operable in successive time-slots; a plurality of data processing units interconnected by the network, and arranged to send streams of messages concurrently through the network, each stream comprising messages that occupy shareable resources in the network in a periodically repeating selection of successive time-slots, a period of repetition (P) being the same for all the streams; node circuits in the network arranged to forward the messages along multi-node paths through the network, each particular stream being assigned a respective stream specific path along which the node circuits forward all messages of the particular stream, the node circuits being arranged to decide whether to forward or discard each message dependent on a measure of seniority of the message in its particular stream.