Resource-intensive applications on smart vehicles is posing difficulties to the use of traditional cloud computing for computation offloading in vehicular networks. In particular, the long transmission distance between the vehicles and the cloud center can cause high latency and poor reliability which may degrade application performance and quality of service. As an integration of mobile edge computing and vehicular networks, vehicular edge computing is a promising paradigm that aims to improve vehicular services by performing computation offloading in close proximity to vehicles. In this paper, the task offloading algorithm that efficiently optimizes task delay and computing resource consumption in multi-user, multi-server vehicular edge computing scenarios is studied. The offloading algorithm not only determines where the tasks are performed, but also indicates the execution order of the tasks on the server. In order to reduce the time complexity, this paper proposes a hybrid intelligent optimization algorithm based on partheno genetic algorithm and heuristic rules. Extensive simulations are conducted, and the results show that compared with the baseline algorithms, the proposed algorithm effectively improves the offloading utility of the VEC system and is suitable for task offloading in various situations.
11 January 2021 in IEEE Transactions on Knowledge and Data Engineering
Network data constantly evolves with new network applications and protocols. There is a need for robust techniques to detect anomalous behaviour. Offline models trained with static data lose validity when new variants of traffic emerge. They require retraining but the need for ground truth and lengthy training times make this task challenging. Meanwhile, online models which detect outliers in streaming data are susceptible to the curse of dimensionality and natural variability. Today's anomalies may be tomorrow's new traffic and existing methods do not provide a way to differentiate between them. We propose a framework that makes the most of both approaches: an offline deep learning model extracts features of normal traffic and provides a bias for an online outlier detection model to select data for training. The online model retains its previously learnt knowledge and retrains itself with new data. Online thresholds are updated in a drifting manner and the Mann-Whitney-U test is incorporated to prevent inaccurate updates. We perform analysis on the scores, develop heuristics to detect new traffic and evaluate using three deep learning models and four outlier detection methods on the UNSW-NB15 and CTU-13 datasets. The framework improves upon any individual offline or online models in isolation.
This paper discusses the design of a broadband wireless network infrastructure which itself is a rain measurement platform for applications such as disaster alarm and sudden hazard decision management systems. A sensor testbed is setup which consists of a hybrid broadband wireless network in conjunction with real-time acoustic rain rate point sensors and complementary rain gauges. The testbed simulates the commercial deployment of a line-of-sight wireless backbone (implemented via a 26 GHz line of sight link) and broadband wireless access network at 5 GHz and 2.4 GHz. Combined wireless signal fade, acoustic power and tipping bucket rain rate measurements over a several month span indicate the feasibility of using rain-induced attenuation and fade durations to trigger imminent-hazard alerts.