DFMEA for Edge Compute Offloading: Deadlines, Protocol Conformance and SLA Evidence

Sending a time-critical function to an edge host saves energy and compute on the device. It also brings new failure modes: radio and queue delay, an unreachable MEC host, and protocol endpoints that don't conform to the spec. This guide shows how to model these in the 7-step AIAG & VDA FMEA Handbook (1st ed., 2019) and IEC 60812:2018, using ETSI GS MEC 003, 3GPP TS 23.501/23.548 (Rel-17+) and, for safety-relevant functions, ISO 26262:2018 and ISO 21448:2022.

Local vs. offload decision

In Step 3 (Function Analysis), give each function a hard deadline. Offload only if the worst-case chain (radio p99.9 + transport + queue + inference + return path) still leaves a documented margin below that deadline. If the function is safety-relevant and the link isn't deterministic, require a local degraded mode and record it as a Prevention Control.

Checklist before you specify

Collect the deadline, p99/p99.9 latency targets, availability SLA, failover time, protocol versions (for example QUIC per RFC 9000, MQTT 5.0 per OASIS) and the conformance test suite to use. Then map them in Step 4: deadline miss as the Failure Mode, stale or missing actuation as the Failure Effect, and congestion, host eviction or a non-conformant peer as Failure Causes. Conformance tests and SLA monitoring become Detection Controls.

Illustrative scenario (analytical, not measured)

A vision inference function has a 100 ms deadline. The assumed budget is 20 ms radio, 15 ms queue, 40 ms inference and 25 ms margin. Preventive action: a local lightweight model that takes over when an 80 ms watchdog fires. Rollout: canary at 5% → 25% → 100%, rolling back automatically if the deadline-miss rate goes above the threshold defined before launch. Re-rate S/O/D and check Action Priority after each stage.

How do I justify the Occurrence rating for an offloaded function in fmeatool.com without inventing data?

Base Occurrence on evidence you can trace: deadline-miss rates from your own pilot or canary telemetry, contractual SLA figures (clearly labeled as commitments, not measurements), and protocol conformance test results. Write the source and the sample conditions (load, radio environment, software version) in the rationale. If you have no evidence yet, use a conservative rating and plan a verification action. Re-rate after the canary stage, and roll back if the observed miss rate goes above the threshold you set in advance.

Back to FMEA Tool · Domains

← Back to FMEA Tool
Engineering & Architecture Brief

DFMEA for Edge Compute Offloading: Deadlines, Protocol Conformance and SLA Evidence

Sending a time-critical function to an edge host saves energy and compute on the device. It also brings new failure modes: radio and queue delay, an unreachable MEC host, and protocol endpoints that don't conform to the spec. This guide shows how to model these in the 7-step AIAG & VDA FMEA Handbook (1st ed., 2019) and IEC 60812:2018, using ETSI GS MEC 003, 3GPP TS 23.501/23.548 (Rel-17+) and, for safety-relevant functions, ISO 26262:2018 and ISO 21448:2022.

Local vs. offload decision

In Step 3 (Function Analysis), give each function a hard deadline. Offload only if the worst-case chain (radio p99.9 + transport + queue + inference + return path) still leaves a documented margin below that deadline. If the function is safety-relevant and the link isn't deterministic, require a local degraded mode and record it as a Prevention Control.

Checklist before you specify

Collect the deadline, p99/p99.9 latency targets, availability SLA, failover time, protocol versions (for example QUIC per RFC 9000, MQTT 5.0 per OASIS) and the conformance test suite to use. Then map them in Step 4: deadline miss as the Failure Mode, stale or missing actuation as the Failure Effect, and congestion, host eviction or a non-conformant peer as Failure Causes. Conformance tests and SLA monitoring become Detection Controls.

Illustrative scenario (analytical, not measured)

A vision inference function has a 100 ms deadline. The assumed budget is 20 ms radio, 15 ms queue, 40 ms inference and 25 ms margin. Preventive action: a local lightweight model that takes over when an 80 ms watchdog fires. Rollout: canary at 5% → 25% → 100%, rolling back automatically if the deadline-miss rate goes above the threshold defined before launch. Re-rate S/O/D and check Action Priority after each stage.

Technical FAQ: How do I justify the Occurrence rating for an offloaded function in fmeatool.com without inventing data?

Base Occurrence on evidence you can trace: deadline-miss rates from your own pilot or canary telemetry, contractual SLA figures (clearly labeled as commitments, not measurements), and protocol conformance test results. Write the source and the sample conditions (load, radio environment, software version) in the rationale. If you have no evidence yet, use a conservative rating and plan a verification action. Re-rate after the canary stage, and roll back if the observed miss rate goes above the threshold you set in advance.

Related Architecture Specifications (FMEATOOL): fmeatools.com · freefmea.com

More Technical Articles on fmeatool.com

Architecture Decision Guide: Local-First DFMEA/PFMEA Execution, Fault-Isolated Storage, and Zero-Trust Data Portability →Timing Failure Modes in DFMEA: Deterministic Telemetry, Sub-Millisecond Latency Budgets and HIL vs. Digital-Twin Detection Controls →FMEA Reference Architecture: Aligning DFMEA/PFMEA Data with AIAG-VDA, SAE J1739 and IEC 60812 →