A Hierarchical Trade-off Assessment Model and the Systematic Evaluation of Networked Systems

In this abstract, a hierarchical trade-off assessment model for networked systems is presented that constructs from the perspectives of systematic objectives, network component requirements, along with multiple quality of service (QoS) dimensions. The primary goal of this conceptual hierarchical trade-off assessment model is to provide a QoS evaluating mechanism for the networked systems. In general, there is no lack of efforts in developing concepts and methods of QoS in networked systems (Lee, 1999; Xiao, 1999; Smirnov, 2001). The model proposed is mainly derived from ISO software quality model (ISO/IEC, 1997). Considerations from network component performance and end users' preferences are taken into account as well. The framework facilitates hierarchical tradeoff through the semantically rich software oriented objective attributes and sub-attributes, technical network components, and QoS metrics measurements. It enables to give guidance on the qualities current networked systems run, and the tradeoffs the networked systems are willing to make under various activities. A series of objective attributes and sub-attributes is borrowed from ISO software quality model. The first level of attributes contains Functionality, Reliability, Availability, Efficiency, Serviceability, and Fairness, selectively adapted from ISO software quality model, other relevant Service Level Agreements (SLA), and end user oriented considerations. The second level deals with apportioning sub-attributes, mapping with six main attributes respectively. To reflect the network level requirements and appropriate impact of multiple network components and services on the overall networked systems characteristics, the third level of the assessment model introduces common network components, such as routers, switches, links, services, and nodes, under each sub-attribute. In practice, the operational characteristics of the networked system are directly measured by the lower performance metrics. Metrics such as delay, jitter, loss, throughput, and utilization are fundamental to performance comparison. Within this theoretical QoS assessment model, these lower level performance metrics are combined with the high level objective and the middle level technical components to explicitly evaluate the actual systemic conditions. Within this multi-dimensional hierarchical structure, partially illustrated in Figure 1, the first three levels of the attributes are normally in non-numeric and nonuniform qualitative order. Even for the lower performance metrics, because many dimensions are involved, it is very difficult to express preferences against each other. In order to formulate a quantitative