Neuromorphic Event-Driven Semantic Communication in Microgrids

TL;DR

This work introduces neuromorphic event-driven semantic communication (NSC) for microgrids, addressing cyber-physical vulnerabilities and scalability in coordination. It deploys spiking neural networks (SNNs) at each converter node and trains them online via Spike-Timing-Dependent Plasticity (STDP) using power-flow measurements, enabling remote-state inference without a dedicated cyber channel. NSC leverages an event-driven data collection scheme and a publish-subscribe architecture to fuse power and information through intrinsic power-flow dynamics, achieving accurate current/voltage and power sharing across diverse microgrid topologies. Simulation and experimental results across two-bus, three-bus, SST-containing, star-topologies and a modified IEEE 14-bus system demonstrate robust performance, energy efficiency, and resilience to line outages and load changes, with scalability for future regional grids.

Abstract

Synergies between advanced communications, computing and artificial intelligence are unraveling new directions of coordinated operation and resiliency in microgrids. On one hand, coordination among sources is facilitated by distributed, privacy-minded processing at multiple locations, whereas on the other hand, it also creates exogenous data arrival paths for adversaries that can lead to cyber-physical attacks amongst other reliability issues in the communication layer. This long-standing problem necessitates new intrinsic ways of exchanging information between converters through power lines to optimize the system's control performance. Going beyond the existing power and data co-transfer technologies that are limited by efficiency and scalability concerns, this paper proposes neuromorphic learning to implant communicative features using spiking neural networks (SNNs) at each node, which is trained collaboratively in an online manner simply using the power exchanges between the nodes. As opposed to the conventional neuromorphic sensors that operate with spiking signals, we employ an event-driven selective process to collect sparse data for training of SNNs. Finally, its multi-fold effectiveness and reliable performance is validated under simulation conditions with different microgrid topologies and components to establish a new direction in the sense-actuate-compute cycle for power electronic dominated grids and microgrids.

Figures (13)

• Figure 1: Going beyond traditional communication norms to task-oriented semantic communications in microgrids: (a) conventional cyber-physical control framework, (b) stages in traditional communication, (c) proposed NSC-based coordinated control framework, (d) simplified and reduced stages in NSC.
• Figure 2: Initial weight determination of SNN at bus $k$ -- Pictorial depiction of a (b) RC circuit and its structural duality with (b) biological neuron, where the input current $I(t)$ acts as an excitation signal into both the circuit and the neuron -- its excitation dynamics are then translated into meaningful output spikes for the learning of NSC using a (c) voltage threshold $V_{th}$ based criteria, (d) Similar to the biological neurons in (b), a converter having disturbances in its input and output can consequently correspond to pre-synaptic and post-synaptic events, respectively, (e) Information-theoretic learning based communication using spiking neural networks (SNNs) and the spike response model (SRM) for simulation of a LIF neuron.
• Figure 3: Online weight update policy of SNN based on the spike-dependent timing plasticity (STDP): (a) Long-term potentiation (LTP) and depression (LTD) based on the excitation of the disturbance observed initially either in the input or the output of the DC/DC converter leading to the trajectory of weight update as per (18), (b) Update of the variables $Q(t)$ and $S(t)$ to formalize the SNN weights as per (24)-(25), (c) update of the synaptic conductance $g_i$ (neuronal) and $g_E$ (total) corresponding to the LTP and LTD events.
• Figure 4: Different topologies for verification: (a) Case I: Two-bus DC microgrid, (b) Case II: Three-bus DC microgrid in ring topology, (c) Case III: Two-bus DC microgrid with an intermediate solid-state transformer, (d) Case IV: Three-bus DC microgrid in star topology.
• Figure 5: Case I: (a) SNN-estimated and sampled currents, (b) SNN-estimated and sampled voltages, (c) captured events, (d) output spikes of SNN, and (e) comparison of SNN against ANN and RNN on power and accumulated energy consumption during a load transient.