Abstract
The risk of soft errors due to radiation continues to be a significant challenge for engineers trying to build systems that can handle harsh environments. Building systems that are Radiation Hardened by Design (RHBD) is the preferred approach, but existing techniques are expensive in terms of performance, power, and/or area. This paper introduces a novel soft-error resilient asynchronous bundled-data design template, SERAD, which uses a combination of temporal and spatial redundancy to mitigate Single Event Transients (SETs) and upsets (SEUs). SERAD uses Error Detecting Logic (EDL) to detect SETs at the inputs of sequential elements and correct them via re-sampling. Because SERAD only pays the delay penalty in the presence of an SET, which rarely occurs, its average performance is comparable to the baseline synchronous design. We tested the SERAD design using a combination of Spice and Verilog simulations and evaluated its impact on area, frequency, and power on an open-core MIPS-like processor using a NCSU 45nm cell library. Our post-synthesis results show that the SERAD design consumes less than half of the area of the Triple Modular Redundancy (TMR), exhibits significantly less performance degradation than Glitch Filtering (GF), and consumes no more total power than the baseline unhardened design.
Introduction
Electronic components in space or military environments are bombarded by high-energy particles, leading to "soft errors." These typically manifest as Single Event Transients (SETs)—temporary voltage glitches in combinational logic—or Single Event Upsets (SEUs)—bit-flips in memory elements like latches or flip-flops.
Traditional Radiation Hardening by Design (RHBD) techniques often impose heavy penalties. Triple Modular Redundancy (TMR) triplicates logic and uses voting, which triples area and power. Glitch Filtering (GF) adds fixed delays to every path to suppress glitches, permanently slowing the system. SERAD addresses this gap by providing a resilient design that maintains near-baseline performance during normal operation, only incurring overhead when an error is actually detected.
The SERAD Architecture
SERAD (Soft Error Resilient Asynchronous Design) leverages an asynchronous bundled-data protocol. Unlike synchronous designs governed by a global clock, SERAD uses local handshaking signals, allowing the system to pause or delay locally without affecting the entire chip.
The architecture consists of several critical components:
- Error Detecting Logic (EDL): Monitors the inputs of sequential elements for glitches (SETs) during the data capture window.
- DICE Latches: Uses Dual Interlocked Storage Cells to provide inherent resistance against SEUs (bit-flips).
- Hardened Control Logic: The asynchronous controller is hardened using Guarded Dual Modular Redundancy (GDMR) and specific gate sizing to prevent control signal corruption.
- Temporal Redundancy: Instead of triplicating logic, SERAD uses re-sampling. If the EDL flags an error, the controller stalls the pipeline and re-samples the data after the glitch has passed.
Performance and Error Recovery
The "pay-as-you-go" performance model is a hallmark of SERAD. Because radiation strikes are relatively rare, the system operates at near-peak frequency most of the time.
As shown in the timing diagram, when an error is detected in Stage 2, the local clock signal is extended to allow for a "Re-sample" phase. This local delay ensures data integrity without requiring a global clock slowdown. This asynchronous flexibility allows SERAD to achieve an average performance nearly identical to unhardened designs.
Evaluation and Results
The researchers implemented SERAD on the "Plasma" open-source MIPS-like processor core using the NCSU 45nm cell library. Performance was measured across standard benchmarks like CoreMark.
- Area Efficiency: SERAD requires only an 80% area overhead, compared to the ~300% overhead typical of TMR.
- Performance: SERAD reached a maximum frequency of 333 MHz, nearly matching the 340 MHz baseline and outperforming the 286 MHz achieved by Glitch Filtering.
- Power Consumption: In arithmetic benchmarks, SERAD actually reduced total power by 23.7% compared to the synchronous baseline. This is attributed to the removal of power-hungry global clock buffers and reduced switching activity in the latch-based design.
Conclusion
SERAD demonstrates that asynchronous design is an ideal framework for soft-error resilience. By moving from "pessimistic" hardening to a reactive, re-sampling approach, engineers can protect systems against radiation without the traditional costs in area and power. While challenges remain in physical design automation and leakage power management, SERAD provides a compelling path forward for high-performance, radiation-hardened electronics.
Summary updated