Montage: Combine Frames with Movement Continuity for Realtime Multi-User Tracking

Abstract

In this work we design and develop Montage for real-time multi-user formation tracking and localization by off-the-shelf smartphones. Montage achieves submeter-level tracking accuracy by integrating temporal and spatial constraints from user movement vector estimation and distance measuring. In Montage we designed a suite of novel techniques to surmount a variety of challenges in real-time tracking, without infrastructure and fingerprints, and without any a priori user-specific (e.g., stride-length and phone-placement) or site-specific (e.g., digitalized map) knowledge: (1) a coded audio tone to support multi-user tracking with minimal latency, in the presence of high noise, multi-path effect and Doppler Shift (2) an innovative stride-length and walking direction estimation method without a priori knowledge of user and site, (3) a vector-based multi-user tracking scheme which connects successive localization snapshots to refine users' locations and generate continuous moving traces. We implemented, deployed and evaluated Montage in both outdoor and indoor environment. Our experimental results (847 traces from 15 users) show that the stride-length estimated by Montage over all users has error within 9cm, and the moving-direction estimated by Montage is within 20 o . For real-time tracking, Montage provides meter-second-level formation tracking accuracy with off-the-shelf mobile phones.

Lasagna

“…Abhinav et al tackle the problem of recognizing smoking behavior using a wristband equipped with a 9-axis inertial sensor [26]. Lan Zhang et al integrate the temporal and spatial constraints while walking and achieve meter-second-level tracking with COTS smartphones [48]. Nirupam Roy et al propose WalkCompass, a system that estimates the walking direction by analyzing the relationship between human walking and its effect on the phone [30].…”

“…Human activity recognition or classification [6,9,26,46,48,50] using smart devices has drawn much attention due to its wide usage and pervasiveness. Those work can be categorized into two groups.…”

“…Those work can be categorized into two groups. The first group is based on physical model, such as walking [48] and smoking [26]. The second group relies on machine learning techniques, which trains classification model for some specific activities using a large amount of data [2,9,10,13,14,50].…”

Indoor acoustic localization: a survey

“…Although it is challenging to achieve high accuracy with inertial sensors, we can still leverage them as an auxiliary tool in acoustic localization. For example, Montage [80] attains an initial position through elapsed time between the two time-ofarrivals which is introduced in detail in "Elapsed time between two time-of-arrivals" section, and updates the position through the movement vector from inertial sensors. CAT [42] leverages inertial sensors to improve the accuracy while plugging their readings into an objective function.…”

“…The acoustic signal is first applied in outdoor localization to detect aircrafts, which is substituted by radars as the RF signal is faster and more effective for long distances. When it comes to indoor localization where GPS does not work well [80], the acoustic signal becomes irreplaceable due to its lower speed, which leads to high accuracy when estimating ToF [59]. Cricket [51] is the first indoor localization system which adopts acoustics and utilizes ToF.…”