A number of recent studies have proposed distributed mutual exclusion algorithms (DMEAs) that coordinate resources in mobile ad-hoc networks (MANETs), vehicular ad-hoc networks (VANETs), and flying ad-hoc networks (FANETs). However, many previous studies focused on performance or security issues in isolation. This paper presents an overview of key DMEAs representing each type of network environment and compares them regarding mechanism class, communication overhead, synchronization delay, response time, fault tolerance, and attack vulnerabilities. A comparison of reported results using quantitative synthesis indicates that representative MANET solutions provide up to 58 O(N^2) to O(N). Resource access mechanisms in FANETs can also be characterized by O(1) best-case message complexity; however, the synchronization times reported vary widely, from T to 4T. These reported synchronization times result from different sets of parameters used during evaluations; therefore, they should be evaluated relatively rather than compared directly. In addition to analyzing performance characteristics, the review also identifies important vulnerabilities related to token manipulation, message interception, impersonation, replay, denial-of-service, and quorum-related attacks that could compromise the safety, liveness, or fairness properties of mutual exclusion. Directions toward mitigating these types of attacks include lightweight authentication, cryptography mechanisms that support privacy preservation, and post-quantum security primitives. Finally, the review points out open problems associated with resource leaks, mobility-based instabilities, context-aware security, and large-scale fault recovery.
Accurately anticipating software faults plays a vital role in improving system reliability and controlling maintenance efforts by highlighting error-prone modules early in the development process. In this work, a stacked deep learning framework combining Bidirectional Long Short-Term Memory units with Gated Recurrent Units is proposed to strengthen the modelling of temporal dependencies within software metrics. The model is thoroughly examined on six established PROMISE datasets, such as JEdit, Ant, Camel, Log4J, Xalan, and Lucene. Across all datasets, the hybrid design consistently outperforms conventional machine-learning approaches and individual recurrent models. It records an average Mean Squared Error of 0.0181, Mean Absolute Error of 0.0614, Average Absolute Error of 0.0579, and an Average Relative Error of 0.0321. The method also yields an average pred(l) accuracy of 98.81
In edge computing environments, efficient task scheduling is critical to improving system utility while balancing task latency and energy consumption. This paper proposes a Bilateral Preference-Aware Task Scheduling Algorithm (BPA) that models mutual preferences between tasks and edge servers to capture both task requirements and resource characteristics. To make effective use of these preferences, BPA applies a two-stage optimization strategy: it first adjusts processor frequency through dynamic voltage and frequency scaling (DVFS), and then refines the scheduling results via task exchanges to further enhance system utility. Extensive simulation results demonstrate that, compared with traditional scheduling algorithms, BPA significantly improves the utilization of system resources. It achieves a well-balanced trade-off between task latency and energy consumption across diverse load conditions, delivering strong overall performance.
Amid the ongoing advancement of healthcare informatization, the sharing of Electronic Medical Records (EMRs) has emerged as a critical issue within the medical and health domain. A key challenge lies in promoting the efficient flow of data while ensuring privacy and security. This paper first proposes an EMR sharing system based on a consortium blockchain architecture and develops an evolutionary game model involving three key stakeholders: patients, medical institutions, and data consumers. The model systematically analyzes the behavioral interactions and strategy evolution of these parties during the processes of EMR authorization, sharing, and usage. In the proposed framework, patients decide whether to authorize the use of their EMRs. Given the patient’s authorization, medical institutions evaluate their own costs and benefits to determine whether to share EMRs, while consumers choose between compliant and non-compliant usage strategies. To enhance system performance, a behavior-based incentive mechanism is introduced, incorporating both rewards and penalties. Subsequently, replicator dynamic equations are constructed to examine the strategy evolution trajectories and identify Evolutionarily Stable Strategies for each party under long-term interactions. Finally, The simulation results show that the proposed mechanism can influence the strategic choices of patients, healthcare institutions, and data consumers, guiding their strategies toward stability and providing a theoretical basis for promoting the stable operation of the EMR sharing system.
Interpolation-based path planning is a popular problem and requires a specific number of data points for curve or line fitting. These points directly affect the solution quality, and it can be argued that determining the optimal number of data points is important. In this paper, a parallel hybrid improved sine cosine-tabu search algorithm (PHISCTS) is proposed, which searches in parallel on a candidate set with different numbers of data points and outputs the number of data points with the shortest path length and its corresponding path. This method combines the strengths of an improved sine cosine algorithm (ISCA) and tabu search (TS). In this method, a hybrid structure was designed in which ISCA and TS algorithms run sequentially. Then, this structure was configured to run in parallel with different numbers of data points on multiple cores. The proposed algorithm was first tested on CEC2015 and CEC2022 benchmark functions and showed superior performance compared to other algorithms. Secondly, it was tested on the path planning problem by designing environments with different difficulty levels. Path planning results demonstrate that the proposed PHISCTS algorithm provides a 2.97
