is a software-only fault detection architecture for safety-critical and embedded systems that eliminates correlated control-flow failures inherent to traditional redundancy techniques. Conventional approaches such as hardware lockstep, Triple Modular Redundancy (TMR), and N-version programming execute identical binaries with identical memory layouts. While effective against many transient faults, these methods remain vulnerable to correlated control-flow errors, where a single program counter perturbation can misdirect all replicas along the same incorrect path, leading to silent data corruption. DME introduces structural address-space decorrelation combined with canonical instruction-trace comparison: • replicas are compiled independently to produce different memory layouts• execution proceeds deterministically in lockstep• each instruction is canonicalized into a layout-independent semantic form• execution results are hashed and compared after every instruction Instead of comparing raw architectural state or physical addresses, the system compares semantic execution traces that include: • opcodes and operands• register identifiers• immediate values• loaded memory values• computed results• stored data Any divergence in control flow or data path produces an immediate hash mismatch, enabling single-instruction latency fault detection. Key properties • software-only (no special hardware required)• detects both control-flow and data corruption faults• protects against correlated faults via layout diversity• deterministic detection latency• low memory footprint (~8 KB RAM for 3 replicas)• suitable for resource-constrained microcontrollers Experimental validation Implementation on ARM Cortex-M microcontrollers demonstrates: • control-flow fault detection ≈ 0.5 µs• data corruption detection within microseconds• minimal runtime overhead Repository contents This archive includes: • full research paper (PDF)• formal model and analysis• implementation details• evaluation results Intended use DME is designed for: • industrial controllers• embedded and real-time systems• safety-critical software• fault-tolerant architectures• research in software-based reliability
Petro Baran (Fri,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: