Hessam Mohammadmoradi, Shengrong Yin, Omprakash Gnawali

Slides:



Advertisements
Similar presentations
Presented by: Richard Wood. Goals and strategies Methods Performance evaluation Performance improvements Remaining Challenges.
Advertisements

FM-BASED INDOOR LOCALIZATION TsungYun 1.
HX-40/40AM 3.0m high mount outdoor detector HX-40: standard model HX-40AM: HX-40 with active IR anti-masking model confidential.
Mohammad Alkhodary Ali Assaihati Supervised by: Dr. Samir Alghadhban EE 578 Simulation Communication Systems Case Study (101) Final.
Doorjamb: Unobtrusive Room-level Tracking of People in Homes using Doorway Sensors Timothy W. Hnat, Erin Griffiths, Ray Dawson, Kamin Whitehouse U of Virginia.
Ultra-Wideband Channel Model for Intra-Vehicular Wireless Sensor Networks C. Umit Bas Electrical and Electronics Engineering, Koc University.
1 Ultrawideband Contents Introduction Why Ultrawideband UWB Specifications Why is UWB unique Data Rates over range How it works UWB Characteristics Advantages.
FLIGHT: Clock Calibration Using Fluorescent Lighting Zhenjiang Li, Wenwei Chen, Cheng Li, Mo Li, Xiang-Yang Li, Yunhao Liu Nanyang Technological University,
Volkan Cevher, Marco F. Duarte, and Richard G. Baraniuk European Signal Processing Conference 2008.
Radio Propagation Spring 07 CS 527 – Lecture 3. Overview Motivation Block diagram of a radio Signal Propagation  Large scale path loss  Small scale.
Watchdog Confident Event Detection in Heterogeneous Sensor Networks Matthew Keally 1, Gang Zhou 1, Guoliang Xing 2 1 College of William and Mary, 2 Michigan.
Advanced Topics in Next- Generation Wireless Networks Qian Zhang Department of Computer Science HKUST Wireless Radio.
Location Systems for Ubiquitous Computing Jeffrey Hightower and Gaetano Borriello.
The National Centres of Competence in Research are managed by the Swiss National Science Foundation on behalf of the Federal Authorities NCCR MICS review.
July 2015 doc.: IEEE /XXXXr0 July 2015
Signal Propagation Propagation: How the Signal are spreading from the receiver to sender. Transmitted to the Receiver in the spherical shape. sender When.
Optimal Power Control, Rate Adaptation and Scheduling for UWB-Based Wireless Networked Control Systems Sinem Coleri Ergen (joint with Yalcin Sadi) Wireless.
Tracking with Unreliable Node Sequences Ziguo Zhong, Ting Zhu, Dan Wang and Tian He Computer Science and Engineering, University of Minnesota Infocom 2009.
Network Computing Laboratory Radio Interferometric Geolocation Miklos Maroti, Peter Volgesi, Sebestyen Dora Branislav Kusy, Gyorgy Balogh, Andras Nadas.
A Power Independent Detection (PID) Method for Ultra Wide Band Impulse Radio Networks Alaeddine EL-FAWAL Joint work with Jean-Yves Le Boudec UWB4SN 2005:
MAC Protocols In Sensor Networks.  MAC allows multiple users to share a common channel.  Conflict-free protocols ensure successful transmission. Channel.
Doc.: IEEE d_Intra-Device_Propagation_Measuremets Submission March 2015 Slide 1 Project: IEEE P Working Group for Wireless Personal.
A Novel one-tap frequency domain RLS equalizer combined with Viterbi decoder using channel state information in OFDM systems Advisor: Yung-an Kao Student:
WINLAB Improving RF-Based Device-Free Passive Localization In Cluttered Indoor Environments Through Probabilistic Classification Methods Rutgers University.
Doc.: IEEE /0553r1 Submission May 2009 Alexander Maltsev, Intel Corp.Slide 1 Path Loss Model Development for TGad Channel Models Date:
RADAR: an In-building RF-based user location and tracking system
Differential Ad Hoc Positioning Systems Presented By: Ramesh Tumati Feb 18, 2004.
Junfeng Xu, Keqiu Li, and Geyong Min IEEE Globecom 2010 Speak: Huei-Rung, Tsai Layered Multi-path Power Control in Underwater Sensor Networks.
1 Research Question  Can a vision-based mobile robot  with limited computation and memory,  and rapidly varying camera positions,  operate autonomously.
CSpy: Finding the Best Quality Channel without Probing Souvik Sen, Bozidar Radunovic, Jeongkeun Lee, Kyn-Han Kim MobiCom’13 1.
Advancing Wireless Link Signatures for Location Distinction Mobicom 2008 Junxing Zhang, Mohammad H. Firooz Neal Patwari, Sneha K. Kasera University of.
Magda El Zarki Ne X tworking’03 June 23-25,2003, Chania, Crete, Greece The First COST-IST(EU)-NSF(USA) Workshop on EXCHANGES & TRENDS IN N ETWORKING 1.
A Power Independent Detection (PID) Method for Ultra Wide Band Impulse Radio Networks Alaeddine EL-FAWAL Joint work with Jean-Yves Le Boudec ICU 2005:
Su-ting, Chuang 1. Outline Introduction Related work Hardware configuration Detection system Optimal parameter estimation framework Conclusion 2.
Spectral Observer with Reduced Information Demand György Orosz, László Sujbert, Gábor Péceli Department of Measurement and Information Systems Budapest.
Doc: IEEE a 5 July 2005 Z. Sahinoglu, Mitsubishi Electric 1 Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs)
Turning a Mobile Device into a Mouse in the Air
CHARACTERIZATION PRESENTATION ANAT KLEMPNER SPRING 2012 SUPERVISED BY: MALISA MARIJAN YONINA ELDAR A Compressed Sensing Based UWB Communication System.
PAPR Reduction Method for OFDM Systems without Side Information
Doc.: IEEE /0632r0 Submission May 2015 Intel CorporationSlide 1 Experimental Measurements for Short Range LOS SU-MIMO Date: Authors:
Accurate WiFi Packet Delivery Rate Estimation and Applications Owais Khan and Lili Qiu. The University of Texas at Austin 1 Infocom 2016, San Francisco.
Shadow Detection in Remotely Sensed Images Based on Self-Adaptive Feature Selection Jiahang Liu, Tao Fang, and Deren Li IEEE TRANSACTIONS ON GEOSCIENCE.
Frame counter: Achieving Accurate and Real-Time Link Estimation in Low Power Wireless Sensor Networks Daibo Liu, Zhichao Cao, Mengshu Hou and Yi Zhang.
Introduction to RPW system
School of Architecture
Teng Wei and Xinyu Zhang
B2W2 N-Way Concurrent Communication for IoT Devices
WP3 INERTIA Local Control and Automation Hub
Application of Digital Inspection in Labeling Systems
Radio Coverage Prediction in Picocell Indoor Networks
A Low Power Carbon Nanotube Chemical Sensor System
KOMUNIKASI DATA Materi Pertemuan 10.
HX-40/40AM 3.0m high mount outdoor detector HX-40: standard model
Walking Speed Detection from 5G Prototype System
Vijay Srinivasan, John Stankovic, Kamin Whitehouse
Ultrawideband Contents
Visible Light based Activity Sensing using Ceiling Photosensors
Full Duplex Benefits and Challenges
Characterizations and Modeling of the Wireless Channel
Bluetooth Based Smart Sensor Network
Hessam Mohammadmoradi Advisor: Prof. Omprakash Gnawali Nov 2017
February 2007 Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [PHY considerations for low power body.
EVM vs PER Plot Not Promising for PSNI
Keystroke Recognition using Wi-Fi Signals
March 2008 Project: IEEE P Working Group for Wireless Personal Area Networks (WPANs) Submission Title: [PHY considerations for low power body area.
RFID Object Localization
Full Duplex Benefits and Challenges
IPSN19 杨景
Proposed TIG on WLAN sensing
Presentation transcript:

Room Occupancy Estimation Through WiFi, UWB, and Light Sensors Mounted on Doorways Hessam Mohammadmoradi, Shengrong Yin, Omprakash Gnawali University of Houston International Conference on Smart Digital Environment (ICSDE ’17) July 21, 2017

Motivation Huge Energy Lost through Buildings 40% of Total Energy Consumed in the US [1] Efficient Management of HVAC 25% Reduction in Energy Waste [2] [1] Buildings Energy Data Book. 2011. Energy Efficiency and Renewable Energy. US department of energy (2011) [2] Alex Beltran, Varick L Erickson, and Alberto E Cerpa. 2013. Thermosense: Occupancy thermal based sensing for hvac control. In Proceedings of the 5th ACM Workshop on Embedded Systems For Energy-Efficient Buildings. ACM, 1–8.

Our Goal Counting people is a key enabler for HVAC control in a smart building Investigate feasibility of using Wireless Sensing to build a people counting solution which is : Inexpensive Reliable Accurate Easy to Mount Privacy Preserving

State of the Art People Counting Solutions Application Cost ($) Privacy Preserving Level Scalability Training Required Mounting Flexibility Break Beam Counting <= 10 High Yes No PIR Presence Ultrasonic <= 100 Moderate RGB Camera Low IR Imager <= 25 Our Solution

Human presence impacts link Approach Count the number of entrances and exits Update occupancy count Put sender and receiver on sides of the door Create an active link between them Human presence impacts link

Channel State Information Channel State Information (CSI) is channel properties of communication link Summarizes impact of environment on signal propagation Scattering Fading Power decay with distance The received signal is compensated using CSI information to build reliable communication link

Utilizing WiFi to Count People IEEE 802.11n defines 64 sub-carriers for 20 MHz bandwidth 128 sub-carriers for 40 MHz bandwidth Sub-carriers face different levels of fading in presence of humans Sub-carrier characteristics can be monitored using CSI information Amplitude and phase information per sub-carrier

Fading levels among WiFi subcarriers in presence of humans Noone between sender and receiver Two persons are standing between sender and receiver Different Subcarriers face different levels of fading in presence of humans

Multiple People between Sender and Receiver One Person standing between sender and receiver Two Persons standing between sender and receiver SNR fluctuation range has a direct relationship with the number of people who are standing in the doorway

Counting People Utilizing WiFi Signals Procedure Count number of people who are walking through the door cnt ← 0 //Number of People cntmax ← 0 // Maximum Number of People CSI ← Channel State Information For sc in CSI sub-carriers do snr ← SNR value for sc snravg ← Average SNR seen for this sub-carrier If |snr − snravg | > α ∗ |snravag | then cnt ← 3 Else if |snr − snravg | > β ∗ |snravg| then cnt ← 2 Else if |snr − snravg| > γ ∗ |snravg| then cnt ← 1 if cntmax < cnt then cntmax ← cnt Return cntmax α = 80% Β = 50% γ = 25%

Utilizing UWB Signals to Count People UltraWideBand wireless signal Bandwidth > 0.2*(Central Frequency) UWB Signal’s Unique Features Resilient to multipath reflection Accurate indoor positioning systems (<50 cm error) Short range, high data rate applications such as room level video streaming Human presence changes the characteristics of the communication channel between sender and receiver

Channel Impulse Response Channel Impulse Response (CIR) is representative of channel characteristics Amplitude and phase information for received signal across time domain (sampling every 1 ns) First Path Signal: The first CIR sample with power significantly higher than background noise

UWB Channel Characteristics in Presence of Humans CIR in presence of two persons vs no person

Counting People Utilizing UWB Signals Procedure Count number of people who are walking though door fppower ← First Path Signal Power noise ← Background Noise cnt ← 0 //Number of People If fppower− noise < α ∗ fppower then cnt ← 0 Else if fppower − noise < β ∗ fppower then cnt ← 1 Else if fppower− noise < γ ∗ fppower then cnt ← 2 Else if fppower − noise < φ ∗ fppower then cnt ← 3 return cnt α = 75% Β = 60% γ = 30% φ = 20% We are interested in instant changes of first path signal compared to background noise

Ambient light level with or without human walking through the door. Illuminance level without presence of human Illuminance level with presence of human

Utilizing Ambient Light Sensing to Count People The key idea is monitoring ambient light’s luminance level and use changes in the level of light due to human presence as an indicator for presence of people Just one sensor is enough to detect presence of people Illuminance level collected when subjects are walking through the door 18 walks through the door

Light Sensing based People Counting Procedure: Detect Presence of People lumavg ← Average Luminance Level in Empty Background lum ← Measured Luminance Level (Window Average) presence ← False // Is Someone There? if lum − lumavg > α ∗ lumavg then presence ← True else presence ← False return presence α = 60%

Evaluation – Experiment Setting TREK1000 Evaluation Kit - UWB Intel 5300 NIC – WiFi Chip TelsoB – Light Sensor WiFi Experiment Setup

Controlled Experiment – Height Flexibility Sender and Receiver nodes mounted at different heights Ten people asked to walk through the door Changes in CSI in different heights (WiFi) Changes in CIR in different heights (UWB) The proposed system can detect the presence of people at deployment heights.

Controlled Experiment – Multiple People Sender and Receiver nodes mounted at the height of 120 cm 1, 2, 3 and 4 people are asked to stand between sender and receiver Changes in CSI with different number of people standing at the Doorway Changes in CIR with different number of people standing at the Doorway Up to 2 people standing side by side are detected by the system

Controlled Experiments – Door Width Door Size 90 cm , 160 cm Deployment Height 120 cm 10 people Accuracy of Counting People inside the Room in a Door with 90 cm Width Event Ground Truth WiFi (%) UWB(%) Light(%) Entrance 55 53 (96%) 54 (98%) 52 (95%) Exit 52 (95%)) 55 (100%) Accuracy of Counting People inside the Room in a Door with 160 cm Width Event Ground Truth WiFi (%) UWB(%) Light(%) Entrance 60 56 (93%) 58 (96%) 54 (86%) Exit 55 (91%)) 59 (98%)

Uncontrolled Experiments Door Size (Conference Room) 100 cm Deployment Height 120 cm Ground Truth Manually counted Accuracy of Counting People inside the Room in Uncontrolled Environment

Uncontrolled Experiment – Combined Solution Door Size (Conference Room) 100 cm Deployment Height 120 cm Ground Truth is counted Manually Combined Solution Majority vote was taken to estimate occupancy inside the room Performance Evaluation in Uncontrolled Environment - Combined Solution Event Ground Truth Combined Solution Entrance 73 70 (96%) Exit 70 68 (97%)

Discussion Light sensing technology is very sensitive to ambient light Rooms with direct sun light are not suitable for light based people counting Maximum detectable crowd size Up to 3 person walking side by side Build more accurate model using machine learning ! All the investigated solutions are very inexpensive Light based people counting solution costs less than 10$

Conclusion Low cost wireless sensing technologies (WiFi, Ultra-wideband and visible light), and others, can be used for people counting The human presence impacts wireless channel Combination of sensing technologies can achieve 96% accuracy in estimating the number of people inside the room Omprakash Gnawali gnawali@cs.uh.edu http://www2.cs.uh.edu/~gnawali/