1. Locating in Fingerprint Space: Wireless Indoor Localization with Little Human Intervention
Jin Yan
Embedded and Pervasive Computing Center
Shanghai Jiao Tong University
Nov 26th, 2015
大家好，今天我给大家介绍一下数据流的一些基本概念和架构。

2. Background & Motivation
<1> Background: Indoor localization is of great importance for a range of pervasive applications.
<2> Motivation: Most radio-based solutions
require a process of site survey, in which radio signatures of an interested area are annotated with their real recorded locations. This causes intensive costs on manpower and time.
Meanwhile, dataflow doesn’t allow any data dependent like shared memory to be used.

3. Related work
Fingerprinting-based techniques: the main idea is to fingerprint the surrounding signatures at every location in the areas and build a fingerprint database. The location is then estimated by mapping the measured fingerprints against the database.
Measured fingerprints
Fingerprint database
Location estimation
Multidimensional scaling (MDS): a set of related statistical techniques often used in information visualization for exploring similarities or dissimilarities in data.
Metrix of node-node distance
MDS
Coordinate to each node

4. Received signal strength(RSS)
Introduction
Using novel sensors integrated in modern mobile phones and leverage user motions to construct the radio map of a floor plan, which is previously obtained only by site survey.
On this basis, we design LiFS, an indoor localization system based on off-the- shelf WiFi infrastructure and mobile phones.
Walking distance
Received signal strength(RSS)

5. LiFS Overview
Puts real locations in a floor plan into a high dimension space, which reflects the walking distance between the points.
Fingerprints are associated with their collecting locations, which means they are labeled with locations.
LiFS take an RSS fingerprint sent by a user as keyword to search the fingerprint database. The best matched item is viewed as the location estimation.

6. Stress-free Floor Plan
Walking distance
𝑑 11 ⋯ 𝑑 1𝑛 ⋮ ⋱ ⋮ 𝑑 𝑛1 ⋯ 𝑑 𝑛𝑛
MDS
The MDS give every sample a new coordinate, with which the Euclidian distance reflects the walking distance in a real floor plan.
3D stress-free floor plan

7. Fingerprint Collection
𝑓= 𝑠 1 , 𝑠 2 ,…… 𝑠 𝑛 ,where 𝑠 𝑖 is the RSS of the ith AP.
Let 𝑑 𝑖𝑗 denotes the distance between the positions of 𝑓 𝑖 and 𝑓 𝑗 , it is measured by the number of footsteps during the movement.
To avoid accumulation of measurement errors, we adopt the individual step counts as the metric of walking distance.

8. Fingerprint Space Construction
The operating phase of LiFS starts when the number of collected fingerprints reaches 10 times of the number of the sample locations in the construction of stress-free floor plans.
Although strides are different from person to person, MDS tolerates measurement errors gracefully, due to its over-determined nature, thus we can use a fixed stride length of a person.
Merge similar fingerprints
Use Floyed-Warshall algorithm to compute all-pair shortest paths
MDS the D matrix to d-dimension

9. Mapping
If all fingerprints correspond with the sample locations in the stress-free floor plan, we are able to label each fingerprint with a real location. Such correspondence comes from the spatial similarity between stress-free floor plan and fingerprint space.

10. Experiments

11. Thank you！
Q & A