Ayon Chakraborty and Samir R. Das WINGS Lab Designing a Cloud-Based Infrastructure for Spectrum Sensing: A Case Study for Indoor Spaces Ayon Chakraborty and Samir R. Das WINGS Lab IEEE DCOSS 2016
Mobile Data Demand Skyrockets Capacity Bottleneck Demand > 3X Capacity! Before we delve deeper into the spectrum sensing infrastructure I will try to point out the importance of wireless spectrum. The plot shows a timeline of average mobile broadband demand and capacity. If you look carefully, by 2016 the demand has already surpassed thrice the capacity. In this situation capacity becomes a bottleneck. In particular wireless network capacity. Mobile Broadband: Average Demand Per User Versus Average Capacity Per User Ref: Rysavy Research WINGS Lab
Managing Network Capacity Capacity Bottleneck Solutions More Cell Sites Improve Backhaul / Core Now lets look at ways that address such capacity problems. The first one is by having more cell sites. Describe each. Mention LTE Shannon limit etc. We will focus on ways that help to find more spectrum or share spectrum resources. Better Utilize Spectrum Improve Spectral Efficiency WINGS Lab
Better Utilize Spectrum (e.g. White Space Channels) Available TV White Space RSS Threshold 6MHz channels FCC opened up this spectrum band for unlicensed use in 2008. WINGS Lab
Finding Available White Spaces Freq1 Freq. White Space 1 Available? 2 Approach 1: Estimate pathloss at (X, Y) using Propagation Models. Freq2 Approach 2: Take spectrum measurements at (X, Y). Location: (X, Y) WINGS Lab
Lost White Space: Causes Propagation models perform poorly indoors and in urban canyon environments. Sometimes too complicated to use indoor models extensively and at scale. WINGS Lab
More White Space is Lost Indoors Most people are indoors 80% of the time and 70% of the spectrum demand comes from indoor locations. Propagation model based techniques seems to lose ≈ 70% of the available white space. [Mobicom’13, CoNEXT’14]. Need for fine-grained spatio-temporal spectrum awareness through intelligent spectrum sensing. WINGS Lab
SpecSense System Scan Spectrum Server Sensing Scheduler Messaging Broker Data Handler Measurements Spectrum Database Analytics Engine Mobile Spectrum Sensors Spectrum Database Web Dashboard WINGS Lab
Mobile Spectrum Sensors Samsung Galaxy S6 Phone Raspberry-Pi Platform WINGS Lab
SpecSense Messaging System: MQTT Useful when: Connectivity is intermittent. Bandwidth is at a premium. Lighter headers. Interact with one or more phone apps. Phone / tablet apps need to send data reliably without requiring code retry logic. Saves > 4% battery per day over HTTP to maintain an open stable connection. WINGS Lab
SpecSense Web Dashboard Live at http://130.245.144.191/ WINGS Lab
Indoor Usecase: Channel Selection WS Ch. #1 WS Ch. #2 Which Channel to Use? WINGS Lab
AP-only Sensing. Poorer REM. REM: Radio Environment Map Ch. #1 available Ch. #2 available WINGS Lab
Client-based Sensing Improves REM Ch. #1 NOT available Ch. #2 NOT available WINGS Lab
More Sensors Reduce Estimation Error WINGS Lab
Impacts Performance Estimation error in the REM leads to erroneous channel selection impacting performance. WINGS Lab
Summary A good amount of wireless spectrum resources are lost due to inefficiencies in signal detection. We built an end-to-end distributed spectrum sensing system to make signal detection more reliable. Using our system we demonstrate how a better channel can be selected by employing spectrum sensing capabilities in client devices. WINGS Lab
Thanks! Interested about the SpecSense system? Measurement Data Deploy your own setup Develop SpecSense APIs Contact: aychakrabort@cs.stonybrook.edu WINGS Lab