Evaluation modeling in performance and resource allocation for residential broadband gateways.

摘要

The growth of the broadband network technology, i.e., digital cables and subscriber lines has made it possible to merge voice, data, audio, and video services into a single pipe from which the broadband gateway can redistribute these services to individual devices such as televisions, telephones, facsimiles, audio systems, and Internet applications connected to residential networks. The study of performance qualities of the broadband gateway is vital for the successful design of the broadband gateway, since it has to distribute a variety of data that can be either sensitive to reliability or susceptible to delay and jitter. Moreover, random traffic characteristics of each continuous media service complicate the analysis of the system and raise the cost of building the system. A high performance system that is too expensive to build would not be desirable. Because of these issues, it is natural to model the performance (performance modeling) of the broadband gateway and the engineering management resources (resource allocation modeling) of the development by applying mathematics.; Besides integrating technologies into the broadband gateway to guarantee quality of service (QoS), the performance study undertaken herein, will introduce the queueing, regression and Markov models. These models are investigated to model traffic generated from multimedia services and to estimate the amount of system resources such as memory, bandwidth and processing power cycles needed for the broadband gateway. Outputs derived from models are used to evaluate the performance of the gateway design, to measure the burst in transmitting data and to predict QoS that the broadband gateway can provide.; For resource allocation modeling, queueing theory, Markov processes, and linear regression are applied to model software management and maintenance during the development and testing phases of a software system. The model helps verify the progress of the testing phase and estimate its cost. Based on results derived from this model, a decision can be made about changes in the workforce early rather than letting the product miss the scheduled deadlines. Moreover, the model helps developers and managers measure such data as product reliability, quality of service, and accurately estimate the number of developers needed for product maintenance.