Research Projects
CmpctArch: A Generic Low Power Architecture for Compact Data Structures in Energy Harvesting Devices [2022]
Small sub-mW sensor devices that rely on energy
harvesting often have limited computation capability to locally
process, query, and update data. Such energy harvesting devices
(EHDs) need to operate under a strict power constraint and
are also extremely cost-sensitive. Based on current costs and
estimated near-term trends, we observe that the price of the
nonvolatile memory component dominates (in ~20 USD devices).
Hence, there is a pressing need to reduce the overall memory
footprint in a data-intensive setting.
This paper is the first to propose a generic hardware archi-
tecture, CmpctArch, for implementing compact data structures
(CDSs) on such devices. They reduce the memory footprint by
up to 3.5X without significantly increasing the overall energy
consumption or time taken (max. additional energy 1.04X and
time 1.18X). The hardware implementations are 160-1200X more
energy-efficient and 280-620X faster than the corresponding
software implementations of CDSs. Our generic template can be
used to instantiate a wide variety of data structures commonly
used in EHD applications.
A Survey on Checkpointing Techniques in Intermittent Systems [2021]
Checkpointing plays a key role in energy harvesting systems. We are performing a detailed study of the various checkpointing approaches proposed by researchers. Along with a theoretical study, we are also comparing these approaches experimentally. There are some interesting questions/details about our studay that we present in a FAQ document, which is available here.
EHDSktch: A Generic Low Power Architecture for Sketching in Energy Harvesting Devices [2020]
Energy harvesting devices (EHDs) are becoming extremely preva-
lent in remote and hazardous environments. They sense the ambient
parameters and compute some statistics on them, which are then
sent to a remote server. Due to the resource-constrained nature of
EHDs, it is challenging to perform exact computations on streaming
data; however, if we are willing to tolerate a slight amount of inaccu-
racy, we can leverage the power of sketching algorithms to provide
quick answers with significantly lower energy consumption.
In this paper, we propose a novel hardware architecture called
EHDSktch - a set of IP blocks that can be used to implement most
of the popular sketching algorithms. We demonstrate an energy
savings of 4-10X and a speedup of more than 10X over state-of-
the-art software implementations. Leveraging the temporal locality
further provides us a performance gain of 3-20% in energy and time
and reduces the on-chip memory requirement by at least 50-75%.
FlexiCheck: An Adaptive Checkpointing Architecture for Energy Harvesting Devices [2019]
With the advent of 5G and M2M architectures,
energy harvesting devices are expected to become far more
prevalent. Such devices harvest energy from ambient sources such
as solar energy or vibration energy (from machines) and use it
for sensing the environmental parameters and further processing
them. Given that the rate of energy consumption is more than the
rate of energy production, it is necessary to frequently halt the
processor and accumulate energy from the environment. During
this period it is mandatory to take a checkpoint to avoid the loss
of data. State of the art algorithms use software based methods
that extensively rely on compiler analyses.
In this work, we provide the first formal model for such
systems, and show that we can arrive at an optimal checkpointing
schedule using a quadratically constrained linear program
(QCLP) solver. Using this as a baseline, we show that
existing algorithms for checkpointing significantly underperform.
Furthermore, we prove and demonstrate that when we have a
relatively constant energy source, a greedy algorithm provides
an optimal solution.
Masters' Project
Scalable and Socially Consistent Systems [2018]
This work aims to handle the system overloading by splitting/merging the objects. Particularly, in social-networking sites like Facebook/Twitter, some posts can be accessed by a large number of users, resulting ina hot spot, which the current systems are not able to handle and hence resulting in enormously high latencyand finally in a system crash. In our system we came up with a model in which the users are responded inapproximately constant time and the system can scale easily according to increase in the number of users.
Other Past Projects
AppSche: Intelligent Scheduler for Approximate Tasks on EHDs [2019]
Developed a scheduling algorithm for approximate tasks on Energy Harvesting Devices. The aim of thescheduler is to minimize the per application error. The system was evaluated on Tejas - A cycle accuratesimulator.
ReXplore- Explore as you research [2019]
A php based web application that basically forms a platform for research collaboration within the department.The main functionalities which have been implemented are profile management, event calendar management,publication repository and recommendations. The application also uses HTML and Mysql.