The proliferation of Internet of Things (IoT) devices in the infrastructure of smart cities has posed cybersecurity risks like never before, which have direct implications on the sustainability and energy consumption of cities. In this paper, a multi-faceted Threat-Resilient Energy-Conscious Security Framework (TRECSF) is introduced that combines intrusion detection methods powered by deep learning, blockchain-driven data integrity verification mechanism, and energy-aware security protocols in smart city ecosystems to achieve their sustainability. The new Hybrid Convolutional Neural Network-Long Short-Term Memory (CNN-LSTM) model is introduced to the proposed architecture, which fulfills the purpose of the study to detect the threat in real time with accuracy of 98.7% and at the same time possesses the ability to execute in a resource-constrained edge device. An Adaptive Energy-Security Optimization (AESO) algorithm that we propose is capable of dynamically adjusting and maintaining the level of security overhead against the level of energy consumption undergoing a 34.2% reductions in power consumption relative to the traditional security systems. The blockchain portion is a consensus mechanism with a lightweight design tailored to IoT settings, which guarantees integrity of the data with a 67% reduced computational power as compared to a conventional Proof-of-Work system. Large-scale simulations conducted on realistic smart city network topologies demonstrate that TRECSF achieves a 45.8% reduction in threat detection latency while ensuring data integrity levels of up to 99.2%. and sustainable energy profiles in a variety of attack scenarios such as Distributed Denial of Service (DDoS), False Data Injection Attacks (FDIA) and Man-in-the-Middle (MitM) attacks. The modular structure of the framework facilitates the seamless integration with the current smart metropolitan infrastructure and facilitates the process of the shift to the carbon-neutral operations of an urban organization.

In the software landscape, understanding component impacts on system reliability is pivotal, especially given the unique complexities of modern software systems. This paper presents a model tailored for software reliability assessment. Our approach introduces the “component influence” to measure a single component’s effect on overall system reliability. Additionally, we adapt a state transition model to cater to the diverse architectures of software systems. Using a discrete-time Markov chain, we predict software reliability. We test our model on an actual software system, finding it notably accurate and superior to existing methods. Our work offers a promising direction for those venturing into software reliability enhancement.