Ph.D. Candidate, CS, University of Alberta ||
Former Alberta Innovates Graduate Fellow ||
Former HLF Young Researcher'22 ||
Former JRF, IIT Kanpur, and IIT Kharagpur ||
With Prof. Frank R. Kschischang, Prof. Gregory W. Wornell, and Prof. Lutz Lampe
With Dr. Marcus Collins
With Prof. Frank R. Kschischang, Prof. Gregory W. Wornell, and Prof. Lutz Lampe
Compression in Delay Tolerant Networks under Limited Storage
Pragmatic Evaluation of Piece-Wise Linear Lossy Compression Schemes
Comparisons of existing PLA schemes can be misleading if no due care is taken to account for the exact representation costs of line segments and their encoding structures. This work addresses it and presents a fair evaluation of the representation cost of competing schemes. Our findings demonstrate that schemes producing fewer line segments do not necessarily achieve better compression ratios once the actual representation cost is taken into account.
PCS: Perpetual Compression Scheme
Existing lossy compression algorithms stop when the storage is filled with compressed data -- no space left to keep more compressed data further. But can we go beyond this limitation? Can we continue compressing data over an infinite horizon (until the next data transmission opportunity) while keeping them in the same space without simply discarding previously compressed data? This work addresses this challenge.
EMFLS: Fixed-Size Lossy Compression Scheme
Can we rethink lossy compression not by asking how to improve the compression ratio within an error bound, but by fixing the storage size and asking: how small can we make the reconstruction error? This work proposes a scheme that fixes the size of the compressed data that will be produced for a given volume of raw data.
Low Complexity Compression for Challenged Wireless Environments
This work investigates low-complexity lossy compression strategies for several resource-constrained and communication-challenged wireless environments, such as long-range, low-power, and low-bitrate networks. We compare three lightweight compression approaches that apply lossy compression only when needed and only to the required extent. To maintain low computational overhead, all three approaches rely on PLA, while differing fundamentally in how the approximation is constructed and optimized. Our results demonstrate that the strategy based on explicit minimization of the maximum reconstruction error consistently outperforms the other approaches, making it a promising solution for highly resource-constrained wireless sensing applications.
Performance Analysis between PLA and PPA-based Lossy Compression in Limited Storage
Do higher-degree polynomials always lead to better compression of time series data? While polynomial-based lossy compression methods often assume that increasing degree improves accuracy, this is not always true under strict storage constraints. In resource-limited embedded sensor systems, higher-degree polynomials perform poorly when minimizing maximum reconstruction error. In contrast, piecewise linear approximation—using simple line segments between data points—often achieves lower error and more reliable performance under the same storage limit.
Design of a SWARM Communication and Networking Architecture for Unmanned Aerial Vehicles
UNet: A Generic and Reliable Multi-UAV Communication and Networking Architecture for Heterogeneous Applications
This work addresses the following key questions in designing a generic multi-UAV communication architecture: how to support diverse networking demands, heterogeneous wireless protocols, real-time data delivery, and concurrent multi-application services.
Interoperability in Internet of Things
SensPnP: Seamless Integration of Heterogeneous Sensors with IoT devices
As IoT applications continue to grow, how can we enable seamless integration of third-party peripherals in devices that were not originally designed for them? In this work, we propose SensPnP, a novel plug-and-play solution combining embedded hardware and firmware to integrate heterogeneous third-party sensors with IoT devices without prior sensor-specific knowledge or Internet dependence. We also present an architecture for a PnP-enabled IoT device supporting multiple embedded peripheral communication protocols.
MSSI: Middleware for Unified Semantic and Syntactic Interoperability in Internet of Things
With the growing demand of IoT, there is a need for seamless and reliable communication between heterogeneous IoT devices and the cyber-world to ensure autonomous control over any application process. More specifically, seamless communication requires interoperability between heterogeneous devices (actors) with different semantics and data formats (syntaxes), which makes it more challenging. We propose a unified middleware to address the issue.
Design of a Sensor Node or IoT Device
WSN/IoT Application in Agriculture
AgriSens: IoT-Based Dynamic Irrigation Scheduling System for Water Management of Irrigated Crops
Different crops have varying water demands across their growth stages. Also, the existing irrigation systems are often not designed from the farmer's perspective. Can we design an irrigation system that efficiently manages crop-specific water needs while remaining low-cost and farmer-friendly? This work addresses the exact question.
AID: A Prototype for Agricultural Intrusion Detection Using WSN
This work proposes a hardware prototype for agricultural intrusion detection using WSN. This simple solution generates alarms at the farmer's house and, at the same time, sends a text message to the farmer's cell phone when an intruder enters the field.
Battery-Less Sensing for IoT
Design of place-and-forget sensor nodes in various IoT applications, especially in DTN. When a mobile node approaches these nodes, the nodes receive power wirelessly from the mobile node, start data collection, and transmit it to the remote server via the mobile node.
Activity Recognition in Wireless Body Area Networks
Sensors-as-Service in Big-Sensor-Cloud
Big-Sensor-Cloud Infrastructure: A Holistic Prototype for Provisioning Sensors-as-a-Service
Can sensor networks be offered as a utility like electricity or water? Can sensors be accessed as easily as cloud services, without owning or managing physical sensor networks? This work proposes a Big-Sensor-Cloud Infrastructure that virtualizes large-scale sensor networks and delivers real-time sensing capabilities as a simple Sensors-as-a-Service (Se-aaS) platform.
Software-Defined Wireless Sensor Networks
Soft-WSN: Software-Defined WSN Management System for IoT Applications
How can IoT applications dynamically control both sensor devices and network behavior? This work introduces a software-defined wireless sensor network architecture that provides centralized, application-aware management of devices and network topology.
Path Planning of Mobile Sinks in Wireless Sensor Networks
Path determination algorithm of Mobile Sinks for energy efficient data collection and optimal coverage in Wireless Sensor Network
What is the best way for mobile data collectors to traverse a sensor network? This work develops a movement strategy for mobile sinks that improves network coverage and energy efficiency while reducing communication costs.
Energy-Cloud in Smart Grid
Cloud-based automated system for on-demand and without service delay supply of energy to end users
Can energy requests in a smart grid be served instantly and optimally without user intervention? This invention introduces a cloud-managed smart grid that automatically allocates power from microgrids to users while minimizing delay and balancing cost.