Ayon Chakraborty

Researcher, NEC Labs America

Responsive Image

I work as a researcher in the Mobile Communications group at NEC Labs America located in Princeton, NJ. Previously, I received my PhD from SUNY Stony Brook where I was advised by Prof. Samir Das. Even earlier, I received a B.E. in Computer Science from Jadavpur University. I hail from Kolkata, India.

Current Research. Scalable indoor/outdoor wireless sensing (RFID, UWB, mm-Waves, acoustics etc.) via autonomous mobile platforms such as robots, drones and so on. Such sensing further enables various micro-services like user or device localization (NSDI'19), warehouse management, surveillance and tactical communications (CoNEXT'18).
Dissertation Research. My dissertation looked at topics related to Dynamic Spectrum Access (INFOCOM'18, INFOCOM'17, HotWireless'16, CoNEXT'14, Mobicom'13), Quality of Experience (CoNEXT'16, IMC'14), Cellular Localization (INFOCOM'15).

TrackIO localization technology selected as a semi-finalist at TEEX UnderFire, Response Innovation Showdown, College Station, TX.
Presented TrackIO at USENIX NSDI 2019, Boston, MA.
Our work (TrackIO) on infrastructure-free tracking of first responders to appear in USENIX NSDI 2019.
Will serve in the program committee for ICDCS 2019 (Wireless track).
Our work on UAV based LTE communication (SkyRAN) to appear in ACM CoNEXT 2018. (acceptance ratio: 17%)
Will be giving an invited talk about our ongoing efforts on UAV based communication infrastructure, SkyLiTE, at the 39th IEEE Sarnoff Symposium 2018. SkyLiTE paper draft, arXiv link.
Our work on spectrum patrolling using crowdsourced spectrum sensors, to appear in IEEE INFOCOM 2018. (acceptance ratio: 19%)
Will serve in the program committee for ICDCS 2018 (Wireless track).
Started working as a researcher at NEC Labs Princeton, NJ.
Successfully defended my Ph.D thesis. Slides, Talk
Invited talk at NEC Labs America, Princeton, NJ.
Invited talk at Indian Institute of Technology - Bombay (IITB). Mumbai, India.
Invited talk at Indian Institute of Technology - Madras (IITM). Chennai, India.
Invited talk at Indian Institute of Technology - Delhi (IITD). New Delhi, India.
Invited talk at Xerox Research Center India (Conduent Labs), Bangalore, India.
Invited talk at Samsung Research America, Mountainview, CA.
Presented Experiential Capacity work at ACM CoNEXT 2016, Irvine, CA.
Our work on large scale distributed RF spectrum monitoring system, SpecSense, to appear in IEEE INFOCOM 2017. (acceptance ratio: 20.9%)
Received ACM travel award and NSF/GENI award for attending ACM Sigcomm CoNEXT 2016, Irvine, CA.
Our proposal on Bringing Spectrum Sensing to the Masses got funded by NSF. The abstract is here.
Our work (with HP Labs) on experience-capacity of wireless networks to appear in ACM Sigcomm CoNEXT 2016. (acceptance ratio: 18%)
I am teaching a graduate course on wireless signals and its applications this Fall.
Two papers in IEEE DCOSS 2016 and ACM HotWireless 2016 on mobile spectrum sensing.
Filed two patents based on my work at HP Labs. Our work is being currently productized by Aruba Networks (acquired by HP) in their admisison control/network selection technologies.
Started working as a research intern in HP Labs (Palo Alto, CA) in the Mobility and Networking group.
Presented Cellular Localization work at IEEE INFOCOM 2015.
Presented Measurement-Augmented spectrum database work at ACM CoNEXT 2014.
Our work on Cellular Localization to appear in IEEE INFOCOM 2015. (acceptance ratio: 19%)
Best paper award nominee at ACM IMC 2014.
Awarded NSF Travel grant for attending ACM Sigcomm CoNEXT 2014, Sydney, Australia.
Awarded NSF Travel grant for attending ACM IMC 2014, Vancouver, Canada.
Our proposal on Measurement-Augmented Spectrum Databases got funded by NSF. The abstract is here.
Our work on White Space Spectrum Databases got accepted in ACM Sigcomm CoNEXT 2014. (acceptance ratio: 19.6%)
Our work on performance comparison of MVNOs to appear in ACM IMC 2014. (acceptance ratio: 22%)
Awarded NSF Travel grant for attending ACM CoNEXT 2013, Santa Barbara, CA.
Presented Mobile spectrum sensing work at ACM MobiCom 2013.
Received NSF Student Travel Award and Microsoft Research ACM SRC grant for the same :)
Adapp accepted in CellNet'13

Ashutosh Dhekne, Ayon Chakraborty, Karthik Sundaresan, Sampath Rangarajan., TrackIO: Tracking First Responders Inside-Out, to appear in USENIX NSDI 2019, Boston, MA (Acceptance Ratio: 12.5%) PDF SLIDES


Ayon Chakraborty, Eugene Chai, Karthik Sundaresan, Amir Khojastepour, Sampath Rangarajan., SkyRAN: A Self-Organizing LTE RAN in the Sky, to appear in ACM CoNEXT 2018, Heraklion, Greece (Acceptance Ratio: 17%) PDF SLIDES


Ayon Chakraborty, Arani Bhattacharya, Snigdha Kamal, Samir Das, Himanshu Gupta and Petar Djuric, Spectrum Patrolling with Crowdsourced Spectrum Sensors, to appear in IEEE INFOCOM 2018, Honolulu, HI. (Acceptance Ratio: 19%) PDF SLIDES


Ayon Chakraborty, Shaifur Rahman, Himanshu Gupta and Samir Das, SpecSense: Crowdsensing for Efficient Querying of Spectrum Occupancy, in IEEE INFOCOM 2017, Atlanta, GA. (Acceptance Ratio: 20.9%) PDF SLIDES


Ayon Chakraborty, Shruti Sanadhya, Samir Das, Dongho Kim and Kyu-Han Kim, ExBox: Experience Management Middlebox for Wireless Networks, in ACM SIGCOMM CoNEXT 2016, Irvine, CA. (Acceptance Ratio: 18%) PDF SLIDES


Ayon Chakraborty, Udit Gupta and Samir Das, Benchmarking Resource Usage for Spectrum Sensing on Commodity Mobile Devices, in ACM HotWireless 2016, New York City. PDF SLIDES


Ayon Chakraborty, Luis Ortiz and Samir Das, Network-side Positioning of Cellular-band Devices with Minimal Effort, in IEEE INFOCOM 2015, Hong Kong. (Acceptance Ratio: 19%) PDF SLIDES


Ayon Chakraborty and Samir Das, Measurement-Augmented Spectrum Databases for White Space Spectrum, in ACM SIGCOMM CoNEXT 2014, Sydney, Australia. (Acceptance Ratio: 19.6%) PDF SLIDES


Fatima Zarinni, Ayon Chakraborty, Vyas Sekar, Phillipa Gill and Samir Das, A First Look at Performance in Mobile Virtual Network Operators, in ACM SIGCOMM IMC 2014, Vancouver, Canada. (Acceptance Ratio: 22.3%) Best paper award nominee PDF SLIDES


Ayon Chakraborty, Samir Das and Milind Buddhikot, Radio Environment Mapping with Mobile Devices in the TV White Space, (Extended Abstract) in ACM MobiCom 2013, Maimi, FLorida. Selected as finalist in ACM Student Research Competetion PDF SLIDES

Dissertation Research: My research deals with two emerging issues in wireless networks. The first issue deals with managing user perceived quality of experience (QoE) in using mobile applications and the second one deals with large scale monitoring of the radio frequency spectrum. My thesis dissertation talk is here.
Responsive Image

  • Quality of Experience [Cellnet'13, IMC'14, CoNEXT'16]: With 2.2M+ apps in the play-store, their diverse requirement for resources and a sky-rocketing user-base, guaranteeing good QoE has become a challenge particularly in resource constrained and highly congested wireless networks. With a multitude of radios available in newer generation smartphones (WiFi, 3G, 4G, LTE) and accessibility to multiple cellular ISPs simultaneously (dual SIMs, Google Fi) a key question we ask is, "Which network is just enough for a good QoE and yet save on the phone's battery, data cost and so on?". We develop a user-feedback based system called Adapp that learns the best network suitable for an app and selects the network accordingly. While Adapp optimizes QoE at an individual user level, it doesn't consider QoE of other users or the network's performance as a whole. We build a system called ExBox (experience middlebox) that uses machine learning to optimize QoE across all users in a network. ExBox also adapts to dynamic changes in the network and is able to perform network selection/admission control with a precision of 0.8 - 0.9. In a sense we develop and redefine the notion of a network's capacity in terms of QoE. The thesis also contributes towards extensive QoE measurement studies across different commercial network providers and hardware vendors. [Most of my QoE work was supported by HP Labs.]

    Responsive Image  . Responsive Image

  • Large-scale RF spectrum monitoring [Mobicom'13, CoNEXT'14, DCOSS'16, HotWireless'16, Infocom'17]: Increasing capacity demands and fixed physical resources in wireless networks bring us to the second issue we focus in this thesis. To cope with such astounding data growth, one viable option is to increase spectrum availability by sharing the spectrum resources across wireless technologies and primary license holders, e.g., cellular providers, broadcast TV etc. The key question we ask here is, "How efficiently can we manage and share available spectrum opportunities across time and space?". Spectrum databases, based on radio-propagation models, help mapping out such availability but are largely erroneous due to inaccuracies in such models. We develop a measurement-augmented spectrum database that subdues such errors by intelligently augmenting the database with real spectrum sensor readings wherever necessary. To make spectrum sensing scalable and pervasive we build spectrum sensing devices of a mobile form-factor (e.g., that runs on Android smartphones). We show via extensive measurement studies that such inexpensive mobile spectrum sensors when present more in number can outperform the measurement accuracy of a few expensive spectrum sensors combined. We build a system called SpecSense that schedules and collects measurements from a distributed system of spectrum sensors in order to estimate spatio-temporal patterns in spectrum availability. Currently we are looking at spectrum surveillance applications running on SpecSense infrastructure to detect illegal transmitters. [Projects funded by NSF, (1) Augmented-DB: link, (2) RF spectrum monitoring: link ]

    Responsive Image  . Responsive Image

  • I have also worked on Cellular Localization [Infocom'15] : We address the problem of network-side localization where cellular operators are interested in localizing cellular devices by means of signal strength measurements alone. While fingerprinting-based approaches have been used recently to address this problem, they require significant amount of geotagged ('labeled') measurement data that is expensive for the operator to collect. Our goal is to use semi-supervised and unsupervised machine learning techniques to reduce or eliminate this effort without compromising the accuracy of localization. Our experimental results in a university campus (6 sq. km) demonstrate that sub-100m median localization accuracy is achievable with very little or no labeled data so long as enough training is possible with 'unlabeled' measurements. This provides an opportunity for the operator to improve the model with time. [This project was supported by Huawei Technologies]

    Responsive Image  . Responsive Image