Showing papers by "Min-You Wu published in 2017"

Folksonomy-Based Internet Object Profiling and Relation Extracting

Abstract: Web 2.0 emphasizes user-generated content, usability, and interoperability for end users. As a killer application in web 2.0, Folksonomy is a collaborative tagging system and enables users to annotate Web objects freely and conveniently. In this paper, we propose an approach to build and enrich profiles of Internet objects (users and resources) based on the relations provided by Folksonomy among tags, resources, and users. This approach extracts semantic relations among tags and constructs a tag hierarchy for calculating similarities of Internet objects, so as to let Internet service or content providers perceive meaning of tags as human beings do. We examine the performance of our approach via cross validation on real datasets and show that, with profile enrichment, recommendation performance measure F1 is increased by 10.0417% compared to recommendation without profile enrichment.

Mrs. Z: Improving ZigBee throughput via multi-rate transmission

Abstract: ZigBee is a widely used wireless technology in low-power and short-range scenarios such as Internet of Things (IoT), sensor networks, and industrial wireless networks. However, the standard ZigBee supports only one data rate, 250Kbps, which thoroughly limits ZigBee's efficiency in dynamic wireless channels. In this paper, we propose Mrs.Z, a novel physical layer design to enable multi-rate selection in ZigBee. The key idea is to change the single spectrum spreading length to multiple ones. Correspondingly, to gracefully adapt to the channel variations, we propose a BER-based rate selection scheme, dividing bit errors into two categories: errors caused by the exceeding despreading threshold, which can be discovered in the physical layer, and caused by incorrect despreading, which is not visible until cyclic redundancy check (CRC) in the media access control (MAC) layer. Then, the receiver selects the rate based on the underlying negative impacts incurred by them and feedbacks to the transceiver. We implement Mrs.Z on USRPs and evaluate its performance in different scenarios. Results demonstrate that Mrs.Z achieves an improvement of 20% and 80% compared to the classic SoftRate and the standard ZigBee.

Classification Statistics in RFID Systems

Abstract: Radio Frequency Identification (RFID) classification statistics problem is defined as classifying the tags into distinct groups and counting the quantity of tags in each group. The issue of time efficiency is significant in classification statistics, especially when the number of tags is large. In such case, the dilemma of short time requirement and massive tags makes traditional one-by-one identification methods impractical. This paper studies the problem of fast classification statistics in RFID systems. To address this problem, we propose a novel Twins Accelerating Gears (TAG) approach. One gear shortens the classification process in frequency domain through subcarrier allocation, when another gear accelerates the statistics process in time domain through geometric distribution based quantity estimation.

On the Profit-Maximizing for Transaction Platforms in Crowd Sensing

Abstract: Crowd sensing is a novel sensing paradigm, in which a challenging task is to balance benefits of various participants, i.e., data requesters, data providers and transaction platforms for attracting sufficient participants. Little attention in literature has been paid to the transaction platform’s profit which is one of the major issues for maintaining a crowd sensing system consistently. In this paper, we aim to propose a mechanism design for optimizing the platform’s profit. For that, we first model the interactions in crowd sensing by leveraging tools of game theory, and then we prove the best strategy for maximizing the benefit of transaction platforms with satisfying individual rationality constraint and incentive compatibility constraint. Finally, we propose two practical algorithms based on the best strategy. Our simulations show that the algorithms are effective in terms of keeping the platform’s profit and time efficiency.