SpecSense: Crowdsensing for Efficient Querying of Spectrum Occupancy

Slides:



Advertisements
Similar presentations
On the Implications of the Log-normal Path Loss Model: An Efficient Method to Deploy and Move Sensor Motes Yin Chen, Andreas Terzis November 2, 2011.
Advertisements

FM-BASED INDOOR LOCALIZATION TsungYun 1.
Comparison of different MIMO-OFDM signal detectors for LTE
David Chu--UC Berkeley Amol Deshpande--University of Maryland Joseph M. Hellerstein--UC Berkeley Intel Research Berkeley Wei Hong--Arched Rock Corp. Approximate.
Topology Control for Effective Interference Cancellation in Multi-User MIMO Networks E. Gelal, K. Pelechrinis, T.S. Kim, I. Broustis Srikanth V. Krishnamurthy,
1 Prediction-based Strategies for Energy Saving in Object Tracking Sensor Networks Yingqi Xu, Wang-Chien Lee Proceedings of the 2004 IEEE International.
Beneficial Caching in Mobile Ad Hoc Networks Bin Tang, Samir Das, Himanshu Gupta Computer Science Department Stony Brook University.
Deterministic Solutions Geostatistical Solutions
Spatial Interpolation
Semi-Supervised Learning Using Randomized Mincuts Avrim Blum, John Lafferty, Raja Reddy, Mugizi Rwebangira.
Deterministic Solutions Geostatistical Solutions
1 Efficient Placement and Dispatch of Sensors in a Wireless Sensor Network Prof. Yu-Chee Tseng Department of Computer Science National Chiao-Tung University.
Jana van Greunen - 228a1 Analysis of Localization Algorithms for Sensor Networks Jana van Greunen.
1 OUTLINE Motivation Distributed Measurements Importance Sampling Results Conclusions.
Ordinary Kriging Process in ArcGIS
Spectrum as a Valuable Resource
Applications in GIS (Kriging Interpolation)
Fast Spectrum Allocation in Coordinated Dynamic Spectrum Access Based Cellular Networks Anand Prabhu Subramanian*, Himanshu Gupta*,
Efficient Gathering of Correlated Data in Sensor Networks
Using ESRI ArcGIS 9.3 Spatial Analyst
$88.65 $ $22.05/A profit increase Improving Wheat Profits Eakly, OK Irrigated, Behind Cotton.
Edoardo PIZZOLI, Chiara PICCINI NTTS New Techniques and Technologies for Statistics SPATIAL DATA REPRESENTATION: AN IMPROVEMENT OF STATISTICAL DISSEMINATION.
Interpolation Tools. Lesson 5 overview  Concepts  Sampling methods  Creating continuous surfaces  Interpolation  Density surfaces in GIS  Interpolators.
Geographic Information Science
Location Estimation in Ad-Hoc Networks with Directional Antennas N. Malhotra M. Krasniewski C. Yang S. Bagchi W. Chappell 5th IEEE International Conference.
GEOSTATISICAL ANALYSIS Course: Special Topics in Remote Sensing & GIS Mirza Muhammad Waqar Contact: EXT:2257.
Spatial Interpolation III
11/25/2015 Wireless Sensor Networks COE 499 Localization Tarek Sheltami KFUPM CCSE COE 1.
Performance Study of Localization Techniques in Zigbee Wireless Sensor Networks Ray Holguin Electrical Engineering Major Dr. Hong Huang Advisor.
On the Role of Dataset Complexity in Case-Based Reasoning Derek Bridge UCC Ireland (based on work done with Lisa Cummins)
Efficient Computing k-Coverage Paths in Multihop Wireless Sensor Networks XuFei Mao, ShaoJie Tang, and Xiang-Yang Li Dept. of Computer Science, Illinois.
Presented By, Shivvasangari Subramani. 1. Introduction 2. Problem Definition 3. Intuition 4. Experiments 5. Real Time Implementation 6. Future Plans 7.
Grid-based Map Analysis Techniques and Modeling Workshop
Lecture 6: Point Interpolation
Cooperative Location- Sensing for Wireless Networks Authors : Haris Fretzagias Maria Papadopouli Presented by cychen IEEE International Conference on Pervasive.
Model Fusion and its Use in Earth Sciences R. Romero, O. Ochoa, A. A. Velasco, and V. Kreinovich Joint Annual Meeting NSF Division of Human Resource Development.
Special Topics in Geo-Business Data Analysis Week 3 Covering Topic 6 Spatial Interpolation.
Optimal Relay Placement for Indoor Sensor Networks Cuiyao Xue †, Yanmin Zhu †, Lei Ni †, Minglu Li †, Bo Li ‡ † Shanghai Jiao Tong University ‡ HK University.
1 Power Efficient Monitoring Management in Sensor Networks A.Zelikovsky Georgia State joint work with P. BermanPennstate G. Calinescu Illinois IT C. Shah.
- A Maximum Likelihood Approach Vinod Kumar Ramachandran ID:
Measurement-Augmented Spectrum Databases for White Space Spectrum
2010 IEEE Global Telecommunications Conference (GLOBECOM 2010)
Prof. Yu-Chee Tseng Department of Computer Science
Labs Put your name on your labs! Layouts: Site 1 Photos Title Legend
Abstract In this paper, the k-coverage problem is formulated as a decision problem, whose goal is to determine whether every point in the service area.
Ayon Chakraborty and Samir R. Das WINGS Lab
An-Najah National University Mobile Wi-Max Network
Boosted Augmented Naive Bayes. Efficient discriminative learning of
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
Ayon Chakraborty, Udit Gupta and Samir R. Das WINGS Lab
Data Mining K-means Algorithm
Discussion of Duty-Cycled Wake-Up Receivers
Advanced Wireless Networks
DASH Background Server provides multiple qualities of the same video
Introduction to Wireless Sensor Networks
Concept of Power Control in Cellular Communication Channels
Guomei Zhang, Man Chu, Jie Li Personal Ubiquitous Computing 2016
Power Efficient Monitoring Management In Sensor Networks
Pose Estimation for non-cooperative Spacecraft Rendevous using CNN
Spatial interpolation
Avidan Efody, Mentor Graphics Corp.
RADAR: An In-Building RF-based User Location and Tracking System
Maximizing Broadcast Tree Lifetime in Wireless Ad Hoc Networks
3D Localization for Sub-Centimeter Sized Devices
Arani Bhattacharya, Han Chen, Peter Milder, Samir R. Das
Connected Sensor Cover Problem
MIRA, SVM, k-NN Lirong Xia. MIRA, SVM, k-NN Lirong Xia.
IPSN19 杨景
Presentation transcript:

SpecSense: Crowdsensing for Efficient Querying of Spectrum Occupancy Md. Shaifur Rahman, Ayon Charkraborty, Himanshu Gupta and Samir R. Das 2017

Outline Spectrum Occupancy Query Interpolation Sensor Selection System Architecture & Benchmarking Conclusion WINGS Lab

Spectrum Occupancy Query Goal: Build infrastructure to respond to spectrum occupancy query Spectrum occupancy query: Is channel f at location (x, y) in use? Why? To identify spectrum sharing opportunities Can help in spectrum patrolling Deeper understanding of spectrum usage for policy formulation How?: Crowdsensing Enables large-scale monitoring via many low-cost low-power sensors. Incentive mechanism. Goal is to build infra,,,,, to respond to this query WINGS Lab

Spectrum Sensing Trade-off VS. High Accuracy Expensive Sensor ThinkRF Realtime Spectrum Analyzer Low Accuracy Inexpensive Sensor RTL-Dongle connected to Cellphone WINGS Lab

Spectrum Sensing Trade-off Challenge: Cost of large-scale deployment of spectrum sensors For a given budget: Small no. of high-accuracy expensive sensors Or Large no. of low-accuracy inexpensive sensors? WINGS Lab

Estimated signal value at (x, y) High Level Overview Sensor SpecSense System Sensor Selection Interpolation Query Location Label sensors, query…..not occupied => signal values Query for Channel: f Location: (x, y) Estimated signal value at (x, y) WINGS Lab

Outline Spectrum Occupancy Query Interpolation Sensor Selection System Architecture & Benchmarking Conclusion WINGS Lab

Interpolation Techniques Inverse Distance Weighting (IDW): Predicted value = distance-weighted average of neighboring values Ordinary Kriging (OK): Takes into accounts structure of the spatial correlation via a “Variogram” function Predicted value = a weighted average of neighboring values, where the weights come from the variogram Minimizes the prediction variance In this work, we improve OK for our context in two ways: Detrending Partitioning WINGS Lab

Interpolation: Ordinary Kriging Predicted value = weighted average of neighboring values Considers spatial correlation Distance: ~ 1 meter Difference: ~ 1 unit Distance: ~ 1 meter Difference: ~ 1 unit WINGS Lab

Variogram Y (30, 625) 30 15 x13 = 50 3 yij 1 (50, 225) x12 = 20 x23 = 65 2 (20, 100) y12 = (15 – 5)2 = 100 5 y13 = (30 – 15)2 = 225 y23 = (30 – 5)2 = 625 (0, 0) h X WINGS Lab

OK Interpolation Z4 = λ1.Z1+ λ2.Z2+ λ3.Z3 = 15λ1+5λ2+30λ3 Find λ1, λ2, λ3 such that: λ1+λ2+λ3 = 1 Expected prediction error = 0 Expected prediction variance is minimum λ1 λ2 λ3 α = 𝑦11 𝑦12 𝑦13 1 𝑦21 𝑦22 𝑦23 1 𝑦31 𝑦32 𝑦33 1 1 1 1 0 −1 𝑦14 𝑦24 𝑦34 1 Z4 = 15(0.2)+5(0.15)+30(0.65) = 23.25 30 15 3 1 x14 = 50 x34 = 20 4 2 x24 = 45 ? 5 WINGS Lab

Improving OK by Detrending OK requires mean of signal values to be constant across all locations But that is not the case in our context WINGS Lab

Improving OK by Detrending Solution: Decompose sensed signal values Signal value = Path-loss + Shadowing Path-loss Estimation Assume TX at the sensor with max. signal Estimate path loss exponent α at sensors Estimate path-loss at query using α. Has zero-mean Use OK interpolation WINGS Lab

Improving Ordinary Kriging by Partitioning WINGS Lab

Improving Ordinary Kriging by Partitioning Issue with OK: It assumes that “variance” depends only on distance. Not true in our context (i.e. signal variance depends on terrain) Solution: Partition the region into sub-region such that: each sub-region has similar path-loss characteristics Algorithm (OK with Partitioning): Divide the region into small cells Initially, each cell is a separate sub-region Iteratively merge neighboring sub-regions with similar estimated path-loss exponent. WINGS Lab

Results: Kriging Improvement Dataset Sensor: RTL-SDR connected to smartphones Spectrum: AT&T’s LTE Downlink 1500 locations Spanning indoors and outdoors 15000 m^2 area Change axis title WINGS Lab

Outline Spectrum Occupancy Query Interpolation Sensor Selection System Architecture & Benchmarking Conclusion WINGS Lab

Sensor Selection Problem: Given locations of sensors and potential queries, select at most k sensors to minimize “interpolation prediction error” NP-hard. Two algorithms: Nearest Query Cover: In each iteration, for each query, select the closest remaining sensor. Iterative Query Cover: In each iteration, select the sensor that “covers” most number of queries. Here, “coverage” is based on the variogram. Above generalize to weighted versions (e.g., weight may be battery-level, precision etc. ) Diagram to ILLUSTRATE the DIFFERENCE between the two algorithms. WINGS Lab

Results: Sensor Selection Pvg prior work.. WINGS Lab

System Architecture WINGS Lab

Results: Benchmarking Requirements: Sensing + Query Latency has to be small enough to serve in real time Sensing Latency: Very low latency even for high FFT size Query Latency: Negligible compared to sensing latency MOTIVATION: … Sensing Latency Query Latency WINGS Lab

Conclusion Crowdsensing based large scale deployment Address two problems- interpolation & sensor selection Improvement of interpolation by detrending and partitioning Performance comparison on real varied dataset Benchmarking of the implemented system WINGS Lab

End of Presentation WINGS Lab