EVALUATION OF DETERMINISTIC ETHERNET FOR THE MODULAR ACTIVE PROTECTION SYSTEMS FRAMEWORK

The U.S. Army Tank-Automotive Research, Development and Engineering Center (TARDEC) contracted DornerWorks Ltd. to evaluate Ethernet-based networking protocols for the safety-critical RDECOM Modular Active Protection Systems (MAPS) framework (MAF). The MAF requires a universal and robust high-speed communication network that can transmit heterogeneous data at near gigabit speeds in a deterministic fashion with bounded and predictable latency. The objectives were to evaluate candidate protocols through rigorous stressing scenarios to: 1) assess and estimate upper bound of performance including data throughput and reliability; and, 2) detect and identify causes and conditions of data loss or corruption. We assessed four protocols: SAE AS6802 (TTEthernet; TTE), ARINC664p7 (rate-constrained; RC), COTS UDP integrated with these two protocols (best-effort; BE), and UDP on a COTS network under three levels of network saturation and with varying payload sizes. On an unsaturated network, TTE had the greatest one-way latency (≥595.74µs) and variance of latency (s.d. ≥ 275.56 µs) from the applications’ perspectives due to lack of synchronization between the network cycle time and the application transmittal period; RC and BE had low and nearly identical latencies (40.23- 452.09 µs) that increased strongly with payload size; and COTS UDP had, by far, the lowest latency (≤92.69) and was relatively unaffected by payload size. On a saturated network, TTE latency increased by 17.3-44.2% , matching the configured virtual link period; RC latency also increased slightly; BE latency increased substantially (>21 ms); UDP latency approximately doubled for larger payload sizes. Surprisingly, with 100% saturation on a full-duplex switched COTS network, COTS UDP had no lost packets except those originating from the saturating devices..