Yongcai Wang

WarpMap: Accurate and Efficient Indoor Location by Dynamic Warping in Sequence-Type Radio-Map

Radio-map based method has been widely used for indoor location and navigation, but remaining key challenges are: 1) laborious efforts to calibrate a fine-grained radio-map, and 2) the locating result inaccuracy and not robust problems due to random signal strength (RSS) noises. An efficient way to overcome these problems is to collect RSS signatures along indoor paths and utilize sequence matching to enhance the location robustness. But, due to problems of indoor path combinational explosion, random RSS loss during movement, and moving speed disparity during online and offline phases, how to exploit sequence matching in radio-map remains difficult. This paper proposes WarpMap, an efficient sequence-type radio-map model and an accurate indoor location method by dynamic warping. Its distinct features include: 1) an undirected graph model (Trace-graph) for efficiently calibrating and storing sequence-type radio-map, which overcomes the path combinational explosion and RSS miss-of-detection problems; 2) an efficient sub-sequence dynamic time warping (SDTW) algorithm for accurate and efficient on-line locating. We show SDTW can tolerate random RSS disparities at discrete points and handle the moving speed differences in on-line and off-line phases. The impacts of different warping distance functions, RSS preprocessing techniques were also investigated. Extensive experiments in office environments verified the efficiency and accuracy of WarpMap, which can calibrated within ten minutes by one person for 1100 m² area and provides overall nearly 20% accuracy improvements than the state-of-the-art of radio-map method.

Accurate and Efficient Indoor Location by Dynamic Warping in Sequence-Type Radio-map

Radio-map based method has been widely used for indoor location and navigation, but remaining key challenges are: 1) laborious efforts to calibrate a fine-grained radio-map, and 2) the locating result inaccuracy and not robust problems due to random signal strength (RSS) noises. An efficient way to overcome these problems is to collect RSS signatures along indoor paths and utilize sequence matching to enhance the location robustness. But, due to problems of indoor path combinational explosion, random RSS loss during movement, and moving speed disparity during online and offline phases, how to exploit sequence matching in radio-map remains difficult. This paper proposes WarpMap, an efficient sequence-type radio-map model and an accurate indoor location method by dynamic warping. Its distinct features include: 1) an undirected graph model (Trace-graph) for efficiently calibrating and storing sequence-type radio-map, which overcomes the path combinational explosion and RSS miss-of-detection problems; 2) an efficient sub-sequence dynamic time warping (SDTW) algorithm for accurate and efficient on-line locating. We show SDTW can tolerate random RSS disparities at discrete points and handle the moving speed differences in on-line and offline phases. The impacts of different warping distance functions, RSS preprocessing techniques were also investigated. Extensive experiments in office environments verified the efficiency and accuracy of WarpMap, which can calibrated within ten minutes by one person for 1100m2 area and provides overall nearly 20% accuracy improvements than the state-of-the-art of radio-map method.

Hop-Constrained Relay Node Placement in Wireless Sensor Networks

Placing relay nodes in wireless sensor networks is a widely-used approach to construct connected network topology. Previous works mainly focus on the relay node minimization while achieving network connectivity but pay less attention on the path performance guarantee. In this paper we first investigate the hop-constrained relay node placement optimization which aims at using as few relays as possible to construct sensor-to-sink paths meeting the hop constraint given by the end user. We present a heuristic-based algorithm to solve the above optimization problem and evaluate its performance by extensive simulation. The experimental results demonstrate that the efficiency of our designs in comparison with two baselines.

Robust Passive Location in Zero-Calibrated Environment Using Smoothed Ordinal Constraints

Passive locating by capturing radio signal strength (RSS) from mobile phones WiFi probing messages in zerocalibrated environments is a challenging problem, because of 1) the lack of accurate RSS-distance model; 2) measurement noise, and 3) the dynamic environment factors affecting the measurements. This paper investigates the noise features of RSS signals by practical experiments, and investigates whether the feasible region bounded by trustworthy ordinal relationships (TOR) among RSS measurements can help to improve the location accuracy. A constrained non-linear optimization model is proposed to apply the TOR constraints for target localization. A series of methods for smoothed feasible region mergence over time are investigated, including 1) Ordinal constraint fusion in one time-slot (OS); 2) Ordinal constraint intersection of multiple time-slots (IM); 3) Ordinal constraint fusion of multiple time-slots using expansion and kernel (EK). The remarkable accuracy and robustness improvements benefited from using TOR constraints were demonstrated by both simulations and practical experiments compared with state-of-the-art passive locating methods.

Joint User Attributes and Item Category in Factor Models for Rating Prediction

One important problem of recommender system is rating prediction. In this paper, we use the movie rating data from MovieLens as an example to show how to use users’ attributes to improve the accuracy of rating prediction. Through data analysis, we observe that users having similar attributes tend to share more similar preferences and users with a special attribute have their own preferred items. Based on the two observations, we assume that a user’s rating to an item is determined by both the user intrinsic characteristics and the user common characteristics. Using the widely adopted latent factor model for rating prediction, in our proposed solution, we use two kinds of latent factors to model a user: one for the user intrinsic characteristics and the other for the user common characteristics. The latter encodes the influence of users’ attributes which include user age, gender and occupation. On the other hand, we jointly use user attributes or item category information and rating data for calculating similarity of users or items. The similarity calculating results are used in our proposed latent factor model as a regularization term to regularize users or items latent factors gap. Experimental results on MovieLens show that by incorporating users’ attributes influences, much lower prediction error is achieved than the state-of-the-art models. The prediction error is further reduced by incorporating influences from item category popularity and item popularity.