Distributed Data Processing Using Intrinsic Network Latencies
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Data processing systems experience significant delays due to intrinsic latencies in network architecture, particularly in routing arrays, format conversion processes, and data transit times through buffer memories, which are exacerbated by the need for ciphering and deciphering operations in System-on-Chip (SoC) systems.
Innovation Solution
The method involves distributing data processing steps across multiple units in the network, utilizing latency times for repetitive processes like ciphering or deciphering, and sending information about the number of iterations remaining for each unit to optimize data transfer by sharing and adapting the processing load based on transit times.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If data is transmitted through the network with ciphering/deciphering operations, then data security is improved, but global latency increases
Solution Approach 1:
The ciphering/deciphering process is divided into multiple iterative steps distributed across different network units. Each unit performs a portion of the ciphering iterations during data transit, rather than performing all ciphering operations at a single location, thereby utilizing network transit time productively and reducing overall latency.
Solution Approach 2:
Ciphering iterations are performed in advance during data transmission through the network, before the data reaches its final destination. The distributed units execute ciphering steps during the network transit time, so that when data arrives at the receiving unit, the ciphering process is already partially or fully complete, eliminating post-transit processing delays.
2Loss of time
If distributed processing is implemented across multiple network units, then global latency is reduced, but device complexity increases
Solution Approach 1:
Multiple network units are equipped with identical or similar processing capabilities to perform ciphering iterations. Each unit can execute the same ciphering algorithm steps, making them universally capable of performing the distributed processing task. This multi-functionality allows any network unit to participate in the ciphering process without requiring specialized hardware at each location.
Solution Approach 2:
The system uses feedback mechanisms to track the progress of ciphering iterations across distributed units. Information about the current iteration state is maintained and communicated between units, allowing the system to coordinate the distributed ciphering process efficiently and ensure that the total number of iterations required for secure encryption is achieved.
Data Source
AI summary
The present disclosure relates to a method for routing data between a sending unit and a receiving unit linked by a network in a processing system comprising several units, the method comprising steps of routing data in the network between the sending unit and the receiving unit, and of applying a process to the routed data, the process comprising several steps which are applied to the data by different units in the network receiving the data, to use latency times in data routing.