This paper introduces some basic concepts of distributed simulation and discusses several techniques for deadlock avoidance and deadlock detection.
The paper starts with an overview of system simulation. The introduction to the basic concepts of modeling and simulation is very poor. Under the section heading of “Traditional Approach to System Simulation,” the author discusses the two commonly used Time Flow Mechanisms (TFMs)--namely, variable-time increment (event-driven) TFM and fixed-time increment (time-driven) TFM. The 12 future event set algorithms [1], the time and state relationships [2], and other relevant literature are completely ignored by the author.
In Section 2, eight good examples are given to illustrate the characteristics of sequential simulation. An obvious fact known to simulationists since the late 1960s is presented as a theorem with a trivial proof. In Section 3, the author presents a model of asynchronous distributed computation, a basic scheme for distributed simulation, and partial correctness and features of the basic distributed simulation scheme. Section 4 discusses the issue of deadlock resolution and presents an overview of the work done by the author, his colleague, and his students.
The paper contains very few references. Out of the 26 given, 11 describe the work done at the author’s university, eight describe other related work, and seven refer to simulation languages and Ada.
All together, the paper contains five theorems and proofs. It does not provide an appropriate survey, nor is it sufficiently tutorial. Why it was published in Computing Surveys, the Survey and Tutorial Journal of the ACM, is a mystery to this reviewer. This paper is the proof that Computing Surveys has lost its survey and tutorial nature.