Massive data streams from IoT and cyber-physical systems must be processed under strict bandwidth, latency, and resource constraints. Generalized Deduplication (GD) is a promising lossless compression framework, as it supports random access and direct analytics on compressed data. However, existing GD algorithms exhibit quadratic complexity $\mathcal{O}(nd^{2})$, which limits their scalability for high-dimensional datasets. This paper proposes \textbf{EntroGD}, an entropy-guided GD framework that decouples analytical fidelity from compression efficiency to achieve linear complexity $\mathcal{O}(nd)$. EntroGD adopts a two-stage design, first constructing compact condensed samples to preserve information critical for analytics, and then applying entropy-based bit selection to maximize compression. Experiments on 18 IoT datasets show that EntroGD reduces configuration time by up to $53.5\times$ compared to state-of-the-art GD compressors. Moreover, by enabling analytics with access to only $2.6\%$ of the original data volume, EntroGD accelerates clustering by up to $31.6\times$ with negligible loss in accuracy. Overall, EntroGD provides a scalable and system-efficient solution for direct analytics on compressed IoT data.

Underwater acoustic signal recognition (UASR) is significant for marine life and ecological environment protection. However, 2-D fixed-parameter inputs are inadequate for adapting to the variable underwater acoustic environment, and learnable-parameter inputs with inductive bias priors in transformers lead to difficulties in model convergence. Additionally, the differing optimization objectives between noise reduction and recognition methods can cause signal distortion, hindering recognition accuracy. To address these issues, this article proposes an innovative end-to-end soft threshold Swin Transformer model based on a multiloss self-distillation training strategy (ESTMST-ST) for robust recognition of weak underwater acoustic targets. Building on the previously proposed time-frequency Swin Transformer (TFST), we design a learnable dual filter module (LDFM) that decomposes underwater acoustic signals in the frequency direction, with parameters obtained through model training. To improve the model’s antinoise performance, we incorporate a soft threshold strategy within TFST to reduce nonstationary interference in underwater acoustic signals. For enhanced robustness and training efficiency, we introduce a self-distillation training strategy with four specific loss functions in selected stage in TFST. Using publicly available datasets, ShipsEar and DeepShip, we conduct three experiments: fixed signal-to-noise ratio (SNR) UASR, multi-SNR UASR, and model generalization ability tests. The experimental results demonstrate that ESTMST-ST achieves superior performance (at least a 1.6 improvement in${F}1$scores and a 2.2 improvement in kappa coefficients) compared to five state-of-the-art methods across two open-source datasets.

Distributed storage systems typically use erasure codes for fault tolerance to reduce storage overhead. However, the data repair process in erasure-coded systems can generate heavy I/O overhead. Existing methods typically increase redundancy to improve repair speed, but this approach results in substantial storage overhead. To enhance repair speed while reducing storage costs, this paper proposes a Machine Learning-based Adaptive Recovery (MLAR) method. Given an application’s access patterns for a file, MLAR uses an adaptive encoding model to calculate the optimal code for each file. When applying fault tolerance with the optimal code, lower-redundancy codes can achieve faster repair speeds than higher-redundancy codes. MLAR employs machine learning to predict file access patterns. Experimental results replaying real-world I/O workloads show that, MLAR reduces storage overhead by 12.8% and recovery time by 23.7%, compared to state-of-the-art methods.

The massive increase and assimilation of Internet of Things (IoT) devices and services imposes new challenges in sensing, communication, and reliable storage of data generated by the IoT. We focus on scenarios where the IoT devices may lose connectivity for long periods of time and can only rely on other IoT devices to store data reliably. This constitutes a problem of distributed storage where lost devices cannot be replaced by others in the network due to the fact that there are no additional devices arriving to the system and that each device has a limited storage capability. Thus, state-of-the-art approaches for distributed storage in data centers are not applicable. We show that optimal policies for data repair, in terms of bandwidth and storage usage, for this novel scenario can be implemented efficiently in real-devices using network coding. We provide a translation from the theoretical results in [1] into an implementation using Raspberry Pi devices.

This paper proposes a low-complexity solution for the data protection problem without newcomer nodes in Internet of Things (IoT) scenarios, i.e., when device losses cannot be replaced by new devices. Application scenarios include environmental monitoring, data collection, and industrial automation. Although the optimal solution and optimization framework have been studied in previous work to minimize the network costs and storage capacity requirements, this paper shows that the optimal solution has a high complexity as the number of devices increases. Given the massive number of IoT devices, we propose a relaxation to the cut capacity constraints that (a) guarantees data recoverability, (b) achieves the minimum network use, and (c) reduces the problem's complexity dramatically. Our numerical results show that the proposed relaxation allows us to change the computational scaling of the problem. More specifically, we show that the time taken to compute the optimal transmission policy with the relaxation for a system with 800 devices is the same as the time it takes the optimal solution to solve the case of 15 devices.

We consider a production system in a general configuration with a new control strategy: the push policy for the part sending and the kanban mechanism for the work-in-process (WIP). The production system is composed of many stations (or workshops) such as an entry station, a set of workstations, a central station, and an exit station. This type of system is modeled as an open queueing network (OQN) in a general configuration with a Markov-type part sending policy and a machine no blocking (MNB) mechanism. The most important performance measures of the production system are the total throughput of the workstations and the total blocking flow of blocked parts sent from the workstations to the central station. This paper discusses an optimization problem with multiple objectives: allocate kanbans to the workstations so as to simultaneously maximize the total throughput and minimize the total blocking flow. Based on a semi-open decomposition approach, several useful properties of the system are characterized. These properties are used to develop a marginal algorithm for the optimization problem. Moreover, a dynamic simulation approach is devised as a tool for evaluating the quality of the solutions obtained by the algorithm. Numerical experiments are provided to demonstrate the efficiency of the algorithm through the simulation approach.

We consider a production system in a general configuration with a new control strategy: the push mechanism for the part transport and the kanban technique for the work-in-process (WIP). The production system is composed of many stations such as an entrance station, a set of work stations, a central station, and an exit station, that are arranged in a general configuration. The push mechanism is followed for transporting a part from a station to a destination station. The kanban technique is adopted for controlling the WIP in a work station. The production system is modeled by a closed queuing network in a general configuration with a Markov part sending mechanism and a machine no blocking (MNB) technique. An optimal part sending policy that maximizes the expected system throughput is formulated into a long run average semi-Markov decision process. Three solution approaches are developed for obtaining optimal or suboptimal solutions. Numerical examples are given to evaluate the quality of the solutions obtained by the solution approaches.