Derivation of service-time bounds of methods in time-triggered message-triggered objects

An essential requirement in real-time distributed computing (RTDC) is to obtain a high degree of assurance on the timeliness of critical actions taken by the systems. Hence, a desirable RTDC software engineering method must allow us to produce RTDC systems fast enough to meet market demands and to confidently determine service-time bounds of the developed RTDC systems.;The Time-triggered Message-triggered Object (TMO) programming scheme is a practical high-level component-based programming model that significantly reduces the amount of labor required in RTDC programming. The TMO, the central element of this scheme, is a syntactically simple and natural but semantically major RTDC extension of the basic object structure. The autonomous-action capability of TMO stems from the time-triggered methods, which are clearly separated from the message-triggered methods whose executions are activated by service request messages from clients. The TMO Support Middleware (TMOSM) provides the execution support mechanisms for TMOs. The software constructs and execution rules defined by the TMO scheme as well as the structuring principles and execution mechanisms used in TMOSM facilitate the analysis of the major factors that contribute to the service times of TMO-based applications.;This dissertation presents a practical approach for deriving high-confidence tight upper bounds for service times of methods in TMOs. The proposed approach represents a fundamental step towards providing guarantees for timely services in large-scale TMO-based applications. A literature search revealed no previously published results in the area of service-time bound analysis of (i) RTDC systems developed using high-level component-based programming models, in general, and (ii) TMO-based applications, in particular.;The approach implements a systematic divide-and-conquer procedure that takes advantage of the features of the TMO scheme and TMOSM mentioned above. The procedure involves: (i) the derivation of tight time bounds for the contributions of individual factors to the service times of methods in TMOs, and (ii) the stepwise integration of those bounds into service-time bounds for the methods. At each step of the procedure a reasonably safe and tight time bound for the considered factor or group of factors is derived through a hybrid method, which combines execution-time measurements and analytically derived loose bounds. Moreover, the bound integration occurs as the time bounds obtained at one step are used to derive the analytical bounds in the following steps.;The proposed approach was evaluated on a simple multimedia distributed application. Considering the results, the approach holds promise to be equally effective with more complex applications.