Zhihan Liu

An Adaptive and Compressive Data Gathering Scheme in Vehicular Sensor Networks

In vehicular sensor networks, probe vehicles can act as mobile sensors to monitor physical world and report to an urban sensing center. However, the distribution of probe vehicles is uneven over space and time. Data redundancy and vacancy are common phenomena for different spatiotemporal positions, which seriously degrade sensing efficiency and accuracy. To address this issue, we propose an adaptive and compressive data gathering scheme (AC-Sense) based on matrix completion theory. The scheme adaptively determines the locations where to obtain samples from so that the principal features of physical world can be captured with a reduced number of probe vehicles. The spatio-temporal correlation between sensor data is exploited to estimate the un-sampled data. Furthermore, we introduce a feedback mechanism to stabilize sensing performance according to the evaluation of data error. We perform extensive experiments based on real taxicab mobility traces and air quality data in Beijing. The experimental results show that the proposed scheme largely improves sensing efficiency while ensuring required data quality.

On Retrieving Moving Objects Gathering Patterns from Trajectory Data via Spatio-temporal Graph

Moving object gathering pattern represents a group event or incident that involves congregation of moving objects, enabling the prediction of anomalies in traffic system. However, effectively and efficiently discovering the specific gathering pattern turns to be a remaining challenging issue since the large number of moving objects will generate high volume of trajectory data. In order to address this issue, we propose a moving object gathering pattern retrieving method that aims to support the retrieving of gathering patterns by using spatio-temporal graph. In this method, firstly we use a density based clustering algorithm (DBScan) to collect the moving object clusters. Then, we maintain a spatio-temporal graph rather than storing the spatial coordinates to obtain the spatio-temporal changes in real time. Finally, a gathering retrieving algorithm is developed by searching the maximal complete graphs which meet the spatio-temporal constraints. To the best of our knowledge, effectiveness and efficiency of the proposed methods are outperformed other methods on both real and large trajectory data.

Optimization Approach to Depot Location in Car Sharing Systems with Big Data

Determining the location of depots of car sharing systems is a fundamental problem in car sharing systems. Existing methods to determine the location of depots mainly use qualitative method and do not take real demand into account. This paper proposes a novel optimization approach to determine the depot location in car sharing systems scientifically. To predict the car sharing demand accurately, we propose a deep learning approach which has been implemented as a stacked auto-encoder (SAE) model at the bottom with a logistic regression layer at the top. The SAE model is employed for unsupervised feature learning, which has been proved to be effective. Meanwhile the spatial and temporal correlations is considered inherently in the prediction model. The results allow us to determine the location of depots scientifically. Experiments on the datasets illustrate that the proposed model for car sharing demand prediction has superior performance.

Location deployment of depots and resource relocation for connected car-sharing systems through mobile edge computing

Mobile edge computing supports the connected cars to ensure real-time, interactive, secured, and distributed services for customers. Connected car-sharing systems, as the promising appliance of connected cars, provide a convenient transportation mode for citizens’ intra-urban commutes. Determining the locations of depots is the primary job in connected car-sharing systems. Existing methods mainly use qualitative method and do not consider spatial–temporal dynamic travel demands. This article proposes a mobile edge computing–based connected car framework which uses normal taxis as connected cars to describe their Global Positioning System trajectory and perform the computing tasks in each mobile edge computing server independently. A spatial–temporal demand coverage approach is developed to optimize the location of depots. This article proposes a deep learning method to predict car-sharing demand constructed by a stacked auto-encoder model and a logistic regression layer. The stacked auto-encoder model is employed for learning the latent spatial and temporal correlation features of demand. A graph-based resource relocation model is proposed to minimize the cost of relocation considering spatio-temporal variation of car-sharing demand. Experiments performed on the large-scale real-world data sets illustrate that our proposed model has superior performance than existing methods.

Space and Time Constrained Data Offloading in Vehicular Networks

Mobile data offloading is a feasible and cost-effective solution to ease the burden of cellular networks. In Internet of Vehicles, however, existing offloading techniques are hardly applicable to the ubiquitous location-dependent services, which impose strict spatiotemporal constraints on content delivery. Particularly, the spatiotemporal constraints cause a phenomenon where the delivery deadlines are different even for the vehicles who subscribe to the same content. To this end, we propose a space and time constrained data offloading scheme (STCDO). The scheme maintains a probability-based contact graph to represent the near-term transmission opportunities between vehicles. Furthermore, a dynamic structure called offloading tree is introduced to evaluate the influence of each vehicle on opportunistic dissemination. Finally, the scheme uses a greedy algorithm to effectively select appropriate vehicles as offloading seeds. We perform extensive experiments based on the real-world map-driven movement model in the ONE simulator. The experimental results show that the proposed scheme largely offloads the overloaded cellular networks while satisfying the spatiotemporal constraints.

Relaying Message and Motivating Collaboration for VANET Data Service

This paper is faced with group sensing problem, where HD map producers motivate private cars to collect data from real world. Group sensing needs vehicles to communicate physically, and drivers to collaborate strategically. First we consider communication module, three VANET-based methods are proposed to achieve inter-vehicle message relaying. Secondly, we consider collaboration module which motivates drivers to be participants, three motivating methods are derived by integrating relaying methods with collaborating strategies. Some combinations of two modules are discussed and classified from centralized or distributed perspective. Finally, we simulate and analyze two modules. The results shows that centralized method could motivate collaboration at a low price, but brings about heavy communication overhead. In contrast, distributed method requires more incentives and less communication overhead than centralized method. Map producers need to make a balance between communication module and collaboration module if they want to improve effectiveness of group sensing.

A Joint Grid Segmentation Based Affinity Propagation Clustering Method for Big Data

Clustering is useful for discovering underlying groups and identifying interesting patterns in scientific data and engineering systems. Affinity propagation (AP) is an effective clustering algorithm which has been successfully applied to broad areas of computer science. To generate high quality clusters, AP iteratively performs information propagation on the full similarity matrix and requires excessive time to exchange messages between data points. This paper proposes a novel AP clustering method based on grid segmentation. The main ideas of our approach are: (1) to partition the data points into multiple non-overlapping sub-sets to simplify representation of huge data points into smaller sub-sets, (2) to construct sparse similarity matrix to decrease the unnecessary message exchanges. Experimental evaluations on large-scale real-world datasets demonstrate our proposed method has superior performance in effectiveness and efficiency.

Anomaly Detection for Internet of Vehicles: A Trust Management Scheme with Affinity Propagation

Anomaly detection is critical for intelligent vehicle (IV) collaboration. Forming clusters/platoons, IVs can work together to accomplish complex jobs that they are unable to perform individually. To improve security and efficiency of Internet of Vehicles, IVs’ anomaly detection has been extensively studied and a number of trust-based approaches have been proposed. However, most of these proposals either pay little attention to leader-based detection algorithm or ignore the utility of networked Roadside-Units (RSUs). In this paper, we introduce a trust-based anomaly detection scheme for IVs, where some malicious or incapable vehicles are existing on roads. The proposed scheme works by allowing IVs to detect abnormal vehicles, communicate with each other, and finally converge to some trustworthy cluster heads (CHs). Periodically, the CHs take responsibility for intracluster trust management. Moreover, the scheme is enhanced with a distributed supervising mechanism and a central reputation arbitrator to assure robustness and fairness in detecting process. The simulation results show that our scheme can achieve a low detection failure rate below 1%, demonstrating its ability to detect and filter the abnormal vehicles.