智慧家庭聯網之 QoS 資源調配的匯聚策略
No Thumbnail Available
Date
2017
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
無中文摘要。
This dissertation recommends a novel rendezvous strategy for QoS pro- visioning with minimum computing cost over intelligent home networks. Although the usual QoS algorithms such as ow control can be adapted for simplifying the bandwidth allocation, the methods are limited only for local services. The proposed algorithm, termed GRNN QoS (Generalized Regression Neural Network for Quality of Service), is able to provide global home network services while requiring minimum computing complexity. The GRNN QoS is a hybrid combination of ow control and path selection. The ow control is adopted for the adjustment and/or allocation of local bandwidth; whereas, the path selection is used for the collection and/or delivery of local network information to the home networks. With GRNN computation, services provided by the home networks are then global optimality. This algorithm is well-suited for intelligent home networks where the fast QoS provisioning and low computing efforts of the small scale home networks are desired. GRNN QoS establishing profiles of user’s various responses to the communication links to different services is the first step for GRNN QoS algorithm. Based on the profiles GRNN QoS can estimate the future feedback to the services from a user. Upon the bandwidth allocations for receiving positive feedback, the algorithm finds a way to minimize the bandwidth usage of the networks. The main advantages of this algorithm are that it quickly adapts to user response and it does not require offline training. This dissertation provides both analytical and numerical analyses to demonstrate the efficacy of said algorithm.
This dissertation recommends a novel rendezvous strategy for QoS pro- visioning with minimum computing cost over intelligent home networks. Although the usual QoS algorithms such as ow control can be adapted for simplifying the bandwidth allocation, the methods are limited only for local services. The proposed algorithm, termed GRNN QoS (Generalized Regression Neural Network for Quality of Service), is able to provide global home network services while requiring minimum computing complexity. The GRNN QoS is a hybrid combination of ow control and path selection. The ow control is adopted for the adjustment and/or allocation of local bandwidth; whereas, the path selection is used for the collection and/or delivery of local network information to the home networks. With GRNN computation, services provided by the home networks are then global optimality. This algorithm is well-suited for intelligent home networks where the fast QoS provisioning and low computing efforts of the small scale home networks are desired. GRNN QoS establishing profiles of user’s various responses to the communication links to different services is the first step for GRNN QoS algorithm. Based on the profiles GRNN QoS can estimate the future feedback to the services from a user. Upon the bandwidth allocations for receiving positive feedback, the algorithm finds a way to minimize the bandwidth usage of the networks. The main advantages of this algorithm are that it quickly adapts to user response and it does not require offline training. This dissertation provides both analytical and numerical analyses to demonstrate the efficacy of said algorithm.
Description
Keywords
QoS, GRNN, Home Network, Quality of Service, Channel allocation, Prediction algorithms, Bandwidth, Home automation, Algorithm design and analysis, Training, Network Access Device, QoS, GRNN, Home Network, Quality of Service, Channel allocation, Prediction algorithms, Bandwidth, Home automation, Algorithm design and analysis, Training, Network Access Device