The Reduction of the Number of Nullmessages in Conservative LP-Simulation Engines

This document deals with techniques used to reduce the number of nullmessages in a distributed discrete{event simulation. Nullmessages are used to avoid deadlocks which can occurr when simulating physical systems (networks) consisting of one or more physical processes. In the brute{force simplest algorithm 10] the number of nullmessages is very high. Reducing the nullmessages number assures improvements to the simulation performance, and many diierent solutions have been proposed for this purpose. The document is composed by a survey where these diierent solutions are presented followed by a comparison between them. A rst section introducing the distributed discrete{event simulation general issues makes the document as more self-contained as possible. 1 Distributed discrete{event simulation This section takes its main contibution by 10], where the distributed approach to discrete{event simulation is described in detail. As explained in the paper, discrete{event simulation have been traditionally faced in a sequential manner. A variable clock holds the time up to which the physical system has been simulated. A data structure, called the event list, maintains a set of messages, with their associated times of transmission, that are scheduled for the future. Each of these messages is guaranteed to be sent at the associated time in the physical system, provided the sender receives no message before this message transmission time. At each step, the message with the smallest associated future time is removed from the event list, and the transmission of the corresponding message in the physical system is simulated. Sending this message may cause other messages to be sent in the future (which then are added to the event list) or cause previuosly scheduled messages to be canceled (which are removed from the event list). The clock is advanced to the time of the message transmission that was just simulated. Since the events (i.e. message transmission) in the physical system are simulated chronologically, and the simulation clock is advanced after simulation of an event to the time of the next event, this form of simulation is called event driven. The algorithm for sequential event{driven simulation can not be easily adapted for concurrent execution on a number of processors, since the event list cannot be eeectively partitioned for such executions. Distributed simulation 4, 5, 2] ooers a radically diierent approach to simulation. Shared data objects of a sequential simulation, namely the clock and the event list, are discarded: no shared variables are considered in this algorithm. In …