Distributed supply chain simulation in a DEVS/CORBA execution environment

Abstract

The emerging electronic commerce and rapidly changing business environments place strong requirements on a next-generation supply-chain analyzer to simulate the flow of goods through the entire supply chain in a timely manner. Such requirements include scalable and efficient model execution and support for flexible future extensibility based on an open industry standard. This paper presents design considerations for a supply chain modeling and simulation environment to execute in a parallel and distributed manner on a DEVS/CORBA run time infrastructure. We recall that DEVS (Discrete Event System Specification) is a sound formal modeling and simulation framework based on generic dynamic systems concepts that can integrate into a parallel and distributed run time infrastructure. CORBA (Common Object Request Broker Architecture) is an open standard that is rapidly gaining universal business acceptance. It can be employed as middleware to support a heterogeneous, networkcentric, distributed computing environment that includes modeling and simulation as well as other business objects. Implementing a distributed supply chain simulator in a DEVS/CORBA execution environment not only may significantly improve execution speed but also may provide advanced supply chain model development capability based on the DEVS modeling and simulation framework.

Performance Analysis of an Optimistic Simulator for CD++

“…• DEVS/CORBA [12]: a runtime infrastructure on top of CORBA to support distributed simulation of DEVS components.…”

Modular HLA RTI services: the GRIDS approach

“…This is based on an increased interest in using distributed simulation technology for non-defense purposes. Work in this area has already attempted to use various middleware including the HLA [12], CORBA [13] and our approach using GRIDS and a thin agent implementing a conservative synchronization protocol (known as a CPADS-TA [8,9]). …”

Parallel simulation techniques for large-scale discrete-event models

“…A number of studies have been devoted to DEVS-based parallel simulations including: DEVS-C++ [71], DEVS/CORBA [72], DEVSCluster [73], DEVS/P2P [75], DEVS/RMI [76], DEVSim++ [77], and P-DEVSim++ [78]. In [79] a distributed simulation strategy for DEVS is presented which combines conservative and risk-free optimistic strategies.…”

“…For instance, the work presented in [71] used a number of benchmarking models to analyze the total execution time and speed up. In [72] the performance of the DEVS/CORBA environment was evaluated by testing a number of Supply Chain Models. The simulator was also tested by conducting a benchmark simulation for a large-scale logistics system [73].…”

“…The LBTS, CMB, and GLM protocols extend the CD++ parallel DEVS/Cell-DEVS framework to allow running the same model with different synchronization mechanisms. The synchronization protocols are separate layers and can be replaced with one another, allowing to run the same model in sequential, 72 optimistic (using the LTW protocol) or conservative (LBTS,GLM,CMB) versions. These three conservative protocols are comparative to the classical P-DEVS protocol of Chow [11] and Zeigler et al [8].…”