A Multifaceted Approach for Improving Wireless Communications for Control of Safety-Critical Tasks

This paper identifies risks associated with noisy communications channels used to control safety-critical tasks in semi-autonomous systems and explores how these risks can be mitigated through a robust communications design and the application of the safety guidelines identified in MIL-HDBK-516. MIL-HDBK-516 identifies the communication channel Bit Error Rate (BER) as one of the standards necessary for determining data integrity and ensuring Safety of Flight (SOF) for military aircraft. BER is typically determined by an analysis of the link budget parameters modeled by a Gaussian noise channel; however, an over-reliance on BER was examined and found to be insufficient to insure reliable wireless communications links and mitigate all safety concerns. A robust communications system that employs a multifaceted approach to safety is then proposed as a means to improve data integrity and ensure continuity of operations by adding layered defenses advocated by MIL-HDBK-516. These layered protections include Forward Error Correction (FEC) codes, Cyclic Redundancy Checks (CRCs), link diversity, automated link routing, boundary range-checking, error resistant interlocks, and methodical cessations of operations. This layered approach addresses the criteria, standards, and compliance methods established by MIL-HDBK-516 for military autonomous aircraft or Unmanned Aerial Vehicle (UAV) to ensure safe operation. Specifically, this layered communication solution to controlling safety-critical tasks in a UAV in a noisy communications channel was found to reduce the Probability of Error (Pe) from greater than 1.00E-6 to a Pe of less than 1.00E-23 with a corresponding reduction of the Risk Hazard Index (RHI) assessment of remote/catastrophic to extremely-improbable/negligible respectively. The principles identified in this paper are particularly applicable for the control of safety-critical tasks in automated systems such as robotic manufacturing, self-driving cars, and autonomous flight where the result of an unmitigated communication error is potentially catastrophic.