This article formally defines Green NFR (GNFR) as a nonfunctional requirement that specifies a measurable, verifiable constraint on the environmental, economic, social, human, or technical sustainability impact of a software-intensive system throughout its lifecycle. Unlike predominantly conceptual prior efforts such as the GreenSoft reference model and sustainable-NFR (SNFR) frameworks, the proposed framework operationalizes GNFR integration through a four-layer pipeline that combines goal-oriented dependency modeling, multicriteria decision analysis, and model-driven sustainability evaluation while capturing interdependencies between sustainability goals and conventional quality attributes. To evaluate its effectiveness, a renewable energy management system (REMS) is used as a case study. Experimental results demonstrate measurable improvements when GNFRs are systematically integrated into system design. In this REMS case study, the architectural alternative selected by the framework exhibited an 18%–25% reduction in operational energy consumption, a 15% improvement in mean time between failures (MTBF), and only a 5%–8% latency overhead under nominal workloads. A sensitivity analysis with ±20% variations in stakeholder preferences confirmed that the top-ranked architecture remained stable in 94% of resampled scenarios. Overall, the proposed framework provides a systematic and quantitatively validated approach for integrating GNFR into architectural decision making, enabling the development of software systems that balance sustainability with traditional quality attributes.
This article investigates the bipartite time-varying formation tracking for the networked heterogeneous robotic systems with the consideration of the predefined-time and singularity avoidance control. In order to address this issue, a predefined-time distributed prescribed-performance (PTDPP) control algorithm is proposed, which is comprised of a novel prescribed-performance predefined-time (PPPT) estimator and an improved nonsingular whole-process predefined-time (NWPPT) controller. The formation trajectory is estimated by the PPPT estimator within a tunable predefined-time and with the constraint for the estimated formation tracking error. Simultaneously, the joint state tracks the joint auxiliary variable with the whole-process predefined-time under the NWPPT controller. And then, by incorporating a first-order continuously differentiable function into the joint auxiliary variable, the state error in the task space will converge to an adjustable residual region within a predefined time, where the state error within this region converges exponentially to the origin. Hence, the consideration of the singularity term arising from the task space is also addressed. The rationality for the PTDPP control algorithm is proven by two theorems, and its effectiveness is validated by applying it to a networked robotic manipulator model. The proposed algorithm could be applied to practical scenarios such as cooperative manipulation in industry or medicine.
This article develops a practical prescribed-time distributed optimization framework for multi-UAV systems subject to composite disturbances and actuator saturation. Using the standard quadrotor position–attitude cascade as the physical plant, the proposed framework connects the distributed optimizer to translational tracking and thrust/attitude command generation, while a practical prescribed-time extended state observer compensates low-frequency disturbances, sliding-mode control rejects bounded high-frequency residuals, and an antiwindup auxiliary system mitigates saturation effects. Regularized higher-order scaling functions assign the observation, tracking, and optimization deadlines while avoiding singular time-varying gains. A three-stage Lyapunov analysis establishes practical prescribed-time convergence of the observation, saturation-compensated translational-tracking, and distributed-optimization errors over an actuator-compatible admissible initial set. The protocol requires no online Hessian inversion. Monte Carlo, prescribed-time-adjustability, scalability, comparative, and ablation studies validate the proposed framework.
Multi-autonomous underwater vehicle (AUV) target pursuit has attracted substantial attention in marine research. To enhance the pursuit efficiency, a novel algorithm named game-theoretic utility tree long short-term memory (GUT-LSTM), which combines LSTM neural network prediction and GUT decision-making, is proposed. The LSTM is employed to predict the target's position. A hybrid algorithm combining particle swarm optimization and grey wolf optimization is introduced to optimize the LSTM's hyperparameters, thereby improving prediction accuracy. Consequently, the GUT allocates tasks to each pursuer autonomous underwater vehicle based on their current positions and the predicted position of the target. Moreover, an obstacle avoidance algorithm, the model-based interfered fluid dynamical system with angle constraints, is utilized to generate collision-free paths. Simulation results in both 2-D and 3-D environments with obstacles ranging from simple to complex demonstrate the feasibility and efficiency of the proposed GUT-LSTM algorithm.
The Social Internet of Vehicles (SIoV) enables flexible data sharing among vehicles, roadside units, and fog nodes due to the fast growth of SIoV. Nonetheless, secure, efficient, and scalable data access in such highly dynamic environments is still a major challenge. Although some searchable encryption schemes offer confidentiality and a fine-grained access control mechanism; however, they rely on static authentication mechanisms with fixed number of cryptographic operations which results in significant computational and communication overhead. This article proposes a mobility-aware attribute-centric authentication (ACA)-based framework for secure data sharing in fog-based SIoV environments. The proposed approach dynamically evaluates vehicle attributes, such as location, mobility, and trust level allowing for context-aware authentication and adaptation of security enforcement. The proposed scheme designs an ultralow complexity construction by combining ACA with the searchability and fog computing of encryption, eliminating redundant cryptographic operations while preserving strong security guarantees. The proposed scheme is proved to be correct formally, and the security analysis shows that it not only provides confidentiality, integrity, authentication, privacy-preserving, and traceability. Moreover, we also perform theoretical cost analysis and OMNeT++-based network simulations for performance evaluation. The results illustrate that the proposed scheme greatly cuts down computational cost, communication overhead, and execution time compared with existing methods like TAKS-DDS, MABKS, and ABERS. In particular, speedup between 25 and 40% for execution time and traffic has been obtained for a range of cases. Therefore, the proposed ACA-based framework provides an effective and scalable solution for secure and efficient data sharing in real-time SIoV environments.
