This information is provided for historical interest only. The STOW RTI development effort officially ended in 1998. However, the RTI implementation resulting from this effort has lived on through other developers and at last report (late 2005), was slated to become an open source implementation. MIT Lincoln Laboratory has no affiliation with this open source effort, so do not contact us about it.
rti-s is implemented in C++ and presents a C++ API to the application writer. This RTI is distributed as a set of statically linked libraries to be linked into the application. rti-s runs in the same thread of control as the federate. It is not yet multi-threaded, but does support multiple federates on a host (in separate processes), or within a single thread.
rti-s provides all of the Data Distribution Management services described in the HLA Interface Specification 1.0 and became the basis for the DDM section of the 1.3 Interface Specification. It also provides the following transport services:
rti-s/A is intended to provide the basic functions required by the various SAF efforts within STOW. Federation Management support is minimal.
An rti-s/A1 release occurred on 13 Nov 96.
rti-s/B is intended to round out services required by the SAF systems as well as the STOW data logger efforts. In addition, rti-s/B provides improved Federation Management services and initial MOM support. The major focus in rti-s/B is improvements in the scaling behavior of the system (number of objects and simulation hosts that can be supported efficiently).
rti-s/C continues to place emphasis on improving performance and scalability of the over all system.
As of 23 Sept 1998, STOW is using the C7F release of RTI-s.
FST-4 was completed in late September 1997. Early results show that at times this 48 hour exercise was supporting in excess of 7,000 HLA objects.
If you have questions, please send us EMail.
MIT Lincoln Laboratory home page | Distributed Simulation Systems
last updated 22 Feb 2006 /jc
The schedule
The current STOW RTI development schedule is as follows:
Results to date
ModSAF Benchmarks
Using the standard ModSAF benchmarks, an rti-s/A(1) based ModSAF achieved excellent performance relative to a DIS-based ModSAF. In the send-only benchmark, the RTI vs. DIS results were roughly equivalent - ranging from 91-100% the number of entities. When running the benchmark with blasters running (creating 400 remote entities for the ModSAF under test to deal with), the RTI version supported roughly twice as many entities as the DIS based benchmark.
Latency and throughput
Some initial test were run to characterize the latency and throughput performance of the rti-s/A(1) system. The test conditions were:
The results were:
Since the rti-s/A(1) system can bundle updates, a second set of data points were obtained.