Computer Science – Performance
Scientific paper
Sep 2010
adsabs.harvard.edu/cgi-bin/nph-data_query?bibcode=2010amos.confe..34j&link_type=abstract
Proceedings of the Advanced Maui Optical and Space Surveillance Technologies Conference, held in Wailea, Maui, Hawaii, September
Computer Science
Performance
Scientific paper
In this paper we describe how and why we chose parallel discrete event simulation (PDES) as the paradigm for modeling the Space Surveillance Network (SSN) in our modeling framework, TESSA (Testbed Environment for Space Situational Awareness). DES is a simulation paradigm appropriate for systems dominated by discontinuous state changes at times that must be calculated dynamically. It is used primarily for complex man-made systems like telecommunications, vehicular traffic, computer networks, economic models etc., although it is also useful for natural systems that are not described by equations, such as particle systems, population dynamics, epidemics, and combat models. It is much less well known than simple time-stepped simulation methods, but has the great advantage of being time scale independent, so that one can freely mix processes that operate at time scales over many orders of magnitude with no runtime performance penalty.
In simulating the SSN we model in some detail: (a) the orbital dynamics of up to 105 objects, (b) their reflective properties, (c) the ground- and space-based sensor systems in the SSN, (d) the recognition of orbiting objects and determination of their orbits, (e) the cueing and scheduling of sensor observations, (f) the 3-d structure of satellites, and (g) the generation of collision debris. TESSA is thus a mixed continuous-discrete model. But because many different types of discrete objects are involved with such a wide variation in time scale (milliseconds for collisions, hours for orbital periods) it is suitably described using discrete events.
The PDES paradigm is surprising and unusual. In any instantaneous runtime snapshot some parts my be far ahead in simulation time while others lag behind, yet the required causal relationships are always maintained and synchronized correctly, exactly as if the simulation were executed sequentially. The TESSA simulator is custom-built, conservatively synchronized, and designed to scale to thousands of nodes.
There are many PDES platforms we might have used, but two requirements led us to build our own. First, the parallel components of our SSN simulation are coded and maintained by separate teams, so TESSA is designed to support transparent coupling and interoperation of separately compiled components written in any of six programming languages. Second, conventional PDES simulators are designed so that while the parallel components run concurrently, each of them is internally sequential, whereas for TESSA we needed to support MPI-based parallelism within each component.
The TESSA simulator is still a work in progress and currently has some significant limitations. The paper describes those as well.
Jefferson David
Leek J.
No associations
LandOfFree
Application of Parallel Discrete Event Simulation to the Space Surveillance Network does not yet have a rating. At this time, there are no reviews or comments for this scientific paper.
If you have personal experience with Application of Parallel Discrete Event Simulation to the Space Surveillance Network, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Application of Parallel Discrete Event Simulation to the Space Surveillance Network will most certainly appreciate the feedback.
Profile ID: LFWR-SCP-O-1555870