Search

A new framework of data assessment and prioritization for real-time prediction of spatial information is presented. The real-time prediction of spatial information is promising for next-generation mobile networks. Recent developments in machine learning technology have enabled prediction of spatial information, which will be quite useful for smart mobility services including navigation, driving assistance, and self-driving. Other key enablers for forming spatial information are image sensors in mobile devices like smartphones and tablets and in vehicles such as cars and drones and real-time cognitive computing like automatic number/license plate recognition systems and object recognition systems. However, since image data collected by mobile devices and vehicles need to be delivered to the server in real time to extract input data for real-time prediction, the uplink transmission speed of mobile networks is a major impediment. This paper proposes a framework of data assessment and prioritization that reduces the uplink traffic volume while maintaining the prediction accuracy of spatial information. In our framework, machine learning is used to estimate the importance of each data element and to predict spatial information under the limitation of available data. A numerical evaluation using an actual vehicle mobility dataset demonstrated the validity of the proposed framework. Two extension schemes in our framework, which use the ensemble of importance scores obtained from multiple feature selection methods, are also presented to improve its robustness against various machine learning and feature selection methods. We discuss the performance of those schemes through numerical evaluation.

Real-time prediction of spatial information such as road-traffic-related information has attracted much attention. Mobile crowdsensing (MCS), in which mobile user devices such as smartphones equipped with sensors work as distributed mobile sensors, is an effective way of collecting sensor data for real-time prediction of spatial information. Since user devices contributing to MCS incur various costs including energy cost and privacy risk, using incentive mechanisms is one approach to compensate for these costs. However, since, in general, the budget for incentive rewarding is limited, rewards should be effectively allocated with considering the contribution of sensor data to the accuracy in real-time prediction of spatial information, which has not been considered in any prior work. This paper presents a scheme to maximize the accuracy of real-time prediction when allocating incentive rewards to user devices. The proposed scheme estimates the contribution of each user device collecting and sending sensor data to the prediction accuracy. Then, the incentive reward received by a user device collecting and sending data increases in proportion to the contribution the data makes to prediction accuracy. Feature selection is used to extract the contribution of each input data point from a prediction model created by machine learning. Evaluation using a real road-traffic-related dataset demonstrated that the proposed scheme works better in terms of prediction accuracy for various cost conditions than a benchmark scheme, Peer Reviewed, Postprint (author's final draft)

Predicting real-time spatial information from data collected by the mobile Internet of Things (IoT) devices is one solution to the social problems related to road traffic. The mobile IoT devices for real-time spatial information prediction generate an extremely high volume of data, making it impossible to collect all of it through mobile networks. Although some previous works have reduced the volume of transmitted data, the prediction accuracy of real-time spatial information is still not ensured. Therefore, this paper proposes an IoT device control system that reduces the amount of transmitted data used as input for real-time prediction while maintaining the prediction accuracy. The main contribution of this paper is that the proposed system controls data transmission from the mobile IoT devices based on the importance of data extracted from the machine learning model used for the prediction. Feature selection has been widely used for extracting the importance of data from the machine learning model. Feature selection methods were also used to reduce communication overhead in distributed learning. Unlike the conventional usage of feature selection methods, the proposed system uses them to control the data transmission of the mobile IoT devices with priority. In this paper, the proposed system is evaluated with a real-world vehicle mobility dataset in two practical scenarios using the random forest model, which is an extensively used machine learning model. The evaluation results show that the proposed system reduces the amount of transmitted input data for real-time prediction while achieving the same level of prediction accuracy as benchmark methods.

Over the last decade, interpersonal communication has attracted more attention from researchers than before. Although the volume of data generated through various communication devices and tools could be enormous, the recent decrease in storage cost enables us to record and store it. The analysis of interpersonal communication is useful to estimate influence in social relationships among people, to detect communities, and to recommend potential friends for users on social networking services. A network graph, which is a mathematical model that represents people as nodes and past opportunities of interpersonal communication as edges, works in such analysis. However, when the capacity of the number of edges recordable in a graph database is limited, or when only a limited number of edges is used for high-speed analysis, it is still unclear which edges should be prioritized and utilized in the analysis. Previous studies suggested that edges in network graphs can be weighted on the basis of the aggregated duration of connections, the number of connections, or the connection time. However, temporal regularity in interpersonal communication has not been well considered in the previous studies. Therefore, in this paper, we propose an edge weighting method for network graphs from interpersonal communication that determines edge weighs on the basis of the scores obtained from the spectral analysis technique. The spectral analysis technique is utilized to numerically deal with temporal regularity and frequency of interpersonal communication. An examination using real records verifies that by using our edge weighting method, link prediction works better under a condition of the limited number of edges usable for the analysis. We also deeply analyze and present the distributions of the frequencies that characterize interpersonal communication.

The confluence of two emerging paradigms, Internet of Things and sharing economy, has encouraged people to share their assets, which could include personal devices, with others. Atypical example of such altruistic device sharing is “tethering”in cellular networks: an owner who uses a smartphone relays data from/to base stations for others who do not have direct connectivity to cellular networks. However, when people share devices, they would be concerned about costs such as battery or bandwidth. Device owners generally want to reduce their costs when they share their devices with someone who is less socially close to them. This is because it was reported that our altruistic behavior has clear correlation with social closeness; the less close someone is to you, the less altruistic actions you take towards that person. Therefore, we propose a system that uses online social relationships to meet device owners' demand for shared-resource management to enable altruistic device sharing. By acquiring and evaluating online social relationships between a device owner and user, the proposed system automatically determines how much resources the user is allowed to use. In this paper, we implemented a prototype system to measure its authentication overhead. Using this actual overhead measured on the prototype system, we conducted a simulation with a large-scale data set of a real social network to verify that: 1) the proposed system limits the resource usage for guest users who are not as close to the device owners and 2) the overhead of the authentication process in the proposed system does not interfere with the resource sharing with guest users who are close to the device owners.