Dynamic Service Selection in Peer-to-Peer Networks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Peer-to-peer networks face challenges in adapting to changing circumstances due to inflexible service selection methods, which can lead to overloading of high-ranked service providers and inefficient resource utilization, especially when network conditions fluctuate.
Innovation Solution
A computer network system that allows devices to select service providers using both deterministic and probabilistic modes based on quality of service data, with a mechanism to monitor and update service quality, enabling rapid adaptation to changing conditions by occasionally favoring untried providers and reducing reliance on top-ranked ones.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If deterministic service selection based on shared rankings is used to select service providers, then service selection efficiency is improved, but service provider overloading occurs
Solution Approach 1:
The patent implements dynamic service selection by allowing devices to switch between deterministic mode (using quality of service data for efficient selection) and probabilistic mode (random selection among available providers). This dynamic approach prevents overloading of top-ranked providers by occasionally selecting alternative providers, thus resolving the contradiction between selection efficiency and provider overload.
2Reliability
If probabilistic service selection is used to avoid provider overloading, then service provider overload is reduced, but service selection efficiency decreases
Solution Approach 1:
The patent applies partial probabilistic selection by using deterministic service selection as the primary method (maintaining efficiency) while introducing probabilistic selection as a supplementary mechanism (preventing overload). The system does not fully adopt probabilistic selection but uses it partially when needed, thus resolving the contradiction between avoiding overload and maintaining efficiency.
3Measurement precision
If quality of service data is shared across the network to improve service selection, then service selection accuracy is improved, but network communication overhead increases
Solution Approach 1:
The patent extracts and shares only essential quality of service data (such as performance metrics and reliability information) rather than complete service provider profiles. This selective data sharing maintains service selection accuracy while reducing network communication overhead by transmitting only the most relevant information.
Data Source
AI summary
A peer-to-peer network operating in accordance with a service-oriented architecture is disclosed. The peers in the network request services from one another and each keep a record of the quality of service they receive from the other peers. The peers can operate in two modes of service provider selection. In a first mode, the selection is so as to favor service providers which have provided the peer with good service in the past. In a second mode, the selection is probabilistic and can therefore select service providers other than those that have provided good service in the past. Each device keeps track of the relative success of adopting the second mode of selection. By occasionally using the second mode of selection and adopting the second mode of selection more frequently should the relative success of adopting the second mode of selection rise, a more rapid adjustment by the network to changing network conditions is enabled. This leads to a better utilization of the resources of the computers of the network than has hitherto been achieved. The invention finds particular application in distributed applications which dynamically select a Web Service to perform a function at run-time.