Intermittent systems powered by harvested energy face significant challenges due to limited memory and fluctuating energy availability. These challenges are further amplified in real-time applications such as environmental monitoring, where multiple DNN (Deep Neural Network) tasks must satisfy strict timing guarantees and maintain high inference accuracy. Existing solutions based on lightweight static or dynamic DNNs suffer from the lack of energy adaptability and/or excessive memory overhead, making them unsuitable for real-time intermittent systems. To address these limitations, we propose LEAF, a novel layer-wise energy-adaptive framework for real-time intermittent systems executing multiple DNN tasks. LEAF consists of three key components: (i) a runtime-reconfigurable DNN model and training methodology that supports multiple latency–accuracy trade-offs without incurring additional memory overhead; (ii) an energy-aware time budget abstraction that integrates intermittent inference behavior into existing timing guarantee frameworks; and (iii) a runtime scheduling mechanism that jointly leverages the model and abstraction to ensure both timing guarantees and high inference accuracy. Experimental results demonstrate that LEAF ensures the timely, energy-aware execution of multiple DNN tasks with higher accuracy compared to existing solutions, validating its effectiveness for real-time intermittent systems.

Edge devices for robotics in hazardous environments, such as rescue drones, navigate complex terrains while transmitting images to remote servers for anomaly detection, including wildfires. However, these devices operate under strict resource constraints, prioritizing operational-critical tasks (e.g., autonomous navigation) while handling image-processing workloads with minimal overhead. Offloading computation to a remote server can alleviate this burden, but unstable network conditions can degrade accuracy and timeliness. To address these challenges, this paper presents a novel offloading framework that balances computational efficiency and accuracy in image-processing tasks. Specifically, it ensures (R1) a minimum accuracy level for individual image-processing tasks associated with different camera sensors and (R2) maximizes the overall image-processing accuracy across all sensors. Our approach builds on an edge-server collaborative image reconstruction architecture, where images are divided into patches and selectively reconstructed. To achieve R1 and R2, we introduce: (i) a hierarchical scheduler that effectively prioritizes patch transmissions under resource constraints and (ii) a feedback mechanism that adapts to network instability, ensuring reliable offloading and inference. Experimental results demonstrate that our framework maintains high accuracy and timely processing, even under network failures.

As the application scope of DNNs executed on microcontroller units (MCUs) extends to time-critical systems, it becomes important to ensure timing guarantees for increasing demand of DNN inferences. To this end, this paper proposes RT-MDM, the first RealTime scheduling framework for Multiple DNN tasks executed on an MCU using external memory. Identifying execution-order dependencies among segmented DNN models and memory requirements for parallel execution subject to the dependencies, we propose (i) a segment-group-based memory management policy that achieves isolated memory usage within a segment group and sharded memory usage across different segment groups, and (ii) an intra-task scheduler specialized for the proposed policy. Implementing RT-MDM on an actual system and optimizing its parameters for DNN segmentation and segment-group mapping, we demonstrate the effectiveness of RT-MDM in accommodating more DNN tasks while providing their timing guarantees.

Although blockchain technology is being increasingly utilized across various fields, the challenge of providing timing guarantees for transactions remains unmet, which is an obstacle in implementing blockchain solutions for time-sensitive applications such as high-frequency trading and real-time payments. In this paper, we propose the first solution to achieve a timing guarantee on blockchain. To this end, we raise and address two issues for timely transactions on a blockchain: (a) architectural support, and (b) real-time scheduling principles specialized for blockchain. For (a), we modify an existing blockchain network, offering an interface to preferentially select the transactions with the earliest deadlines. We then extend the blockchain network to provide the flexibility of the number of generated blocks at a single block time. Under such architectural supports, we achieve (b) with three steps. First, to resolve a discrepancy between a periodic request of a transaction-generating node and the corresponding arrival on a block-generating node, we translate the former into the latter, which eases the modeling of the transaction load imposed on the blockchain network. Second, we derive a schedulability condition of the modeled transaction load, which guarantees no missed deadline for all transactions under a work-conserving deadline-based scheduling policy. Last, we develop a lazy scheduling policy and its condition, which reduces the number of generated blocks without compromising the degree of timing guarantees for the work-conserving policy. By implementing RT-blockchain on top of an existing open-source blockchain project, we demonstrate the effectiveness of the proposed scheduling principles with architectural supports in not only ensuring timely transactions but also reducing the number of generating blocks.