Commit 507f56ed authored by U-RAJESH-SIS\rajesh's avatar U-RAJESH-SIS\rajesh

edited for size

parent 8db425de
\section{Related Work}
\label{sec:relatedwork}
There has been a wide variety of related work in the broad areas of energy harvesting, including WiFi/RF energy harvesting, low-power wearable design and WiFi beamforming.
There has been a wide variety of related work in the broad areas of energy
harvesting, including WiFi/RF energy harvesting, low-power wearable design
and WiFi beamforming.
%We focus on prior work around our two key concepts of
%WiFi-based energy harvesting and indoor localization.
We classify related work into categories related to our two key concepts: WiFi-based energy harvesting and indoor localization.
%Energy harvesting papers \cite{talla2017battery} and WISP platform, some commercial products
%
%Localization/Activity recognition \cite{xiong2013arraytrack} \cite{pu2013whole}
......@@ -13,17 +17,49 @@ We classify related work into categories related to our two key concepts: WiFi-b
%Simultaneous Wireless Information and Power Transfer (SWIPT)
\subsection{Energy Harvesting for Client Devices}
There is a significant body of literature on energy harvesting for batteryless wearable and embedded sensing devices. The most popular modes of energy harvesting are light, kinetic energy and thermal gradients. Ambient and solar lighting provides some of the highest amount of harvested power: over a decade ago, Heliomotes~\cite{lin2005} were designed to support embedded sensors using solar power, whereas Hande et al~\cite{hande2007} demonstrated the feasibility of powering indoor wireless routers (albeit in a non-wearable form factor) using energy harvested from indoor fluorescent lights. Kinetic energy is another popular source for energy harvesting. For example, the EnergyBug prototype~\cite{ryokai14} focused on harvesting kinetic energy, generated by the body movements of children, to power a wrist-mounted wearable device, whereas SolePower~\cite{XXX} developed a shoe insole that harvests kinetic energy (when a person is walking). More recently, the Flicker prototype~\cite{hester2017} presented a platform for rapid prototyping of energy harvesting-based sensors: the software abstractions allow a device to harness the combined capabilities of different energy harvesting modes, including solar, RFID and kinetic energy, in a federated storage infrastructure. Thermal energy harvesting is based on the principle of using temperature gradients between the human body and the surrounding environment to generate an electrical charge. Examples of sensing prototypes include Thermes~\cite{campbell2014}, an embedded sensor which uses such temperature gradients to detect water usage events in buildings and the work by Xu et al~\cite{xu2013}, which demonstrated the ability to harvest energy using the thermal gradient between a shoe's insole and the external ground. All of this work has two broad characteristics: (i) their power generation capability is dependent on certain context (e.g., kinetic harvesters work only when the user exhibits movement), and (ii) when applied to wearable form factors, they tend to perform relatively low-power tasks and do not usually operate power-hungry sensors, such as an accelerometer.
There is significant prior work on energy harvesting for wearable / embedded
devices using light, kinetic energy, thermal gradients etc. Ambient and
solar lighting generally provides the highest amount of harvested power as
demonstrated by Heliomotes~\cite{lin2005} to power embedded devices and
Hande et. al~\cite{hande2007} to power indoor APs. Kinetic energy is another
popular energy harvesting source that can use body movements (e.g.
EnergyBug~\cite{ryokai14}), and walking (e.g. SolePower~\cite{XXX}) to
power ultra low power body sensors. Thermal energy harvesting uses
temperature gradients to generate an electrical charge. For example,
Thermes~\cite{campbell2014} used thermal harvesting to detect water usage
events in buildings while Xu et. al~\cite{xu2013}, used thermal gradients
between a shoe's insole and the external ground. More recent work, such as
Flicker~\cite{hester2017}, provide a platform for rapid prototyping of
energy harvesting-based sensors. Our work is complementary to these prior
methods and can be used to a) power higher power devices, and b) deployed in
environments (e.g. dark warehouses) where other methods would not work.
\subsection{WiFi \& RF harvesting}
Harvesting of wireless transmission power has also been explored, especially in the context of charging RFID tags and devices. Most of the existing solutions for delivering RF power transfer require custom-designed hardware. The WISP platform~\cite{sample2008}, which operates a 16-bit ultra low power microcontroller based exclusively on harvested RF energy, is probably the best known energy harvesting-based embedded sensing platform, and supports the operation of sensors such as ambient light, temperature and orientation. Recent work has specifically looked at the promise of harvesting WiFi-based RF energy for batteryless operation. The PoWiFi system~\cite{talla2015powering} demonstrates the potential of power transmission using the existing Wi-Fi chipsets by modifying the firmware of the access point (AP), and showed that it is practically possible to harvest energy from WiFi transmissions. However, this system does not explore the possibility of beamforming to improve the harvesting energy. Limited additional work has explored the possibility of energy harvesting from beamformed RF transmissions, but only via simulations. Huang et.al~\cite{huang2016performance} explore the operation of a wirelessly-powered communication network (WPCN), by considering a multi-antenna AP, and deriving the optimal energy harvesting time that maximizes the throughput achievable via an energy-harvesting client. Liu et.al.~\cite{liu2014multi} use a similar multi-antenna AP model to maximize a client's throughput in a WPCN environment, by optimally dividing the time between uplink data transmissions and and downlink energy beamforming.
\subsection{WiFi-based Localization \& gesture recognition}
Our proposed \name framework is based on the ability to perform sufficiently accurate tracking of a wearable, potentially mobile, device, so that the WiFi AP can deliver an adequately high quantity of RF energy to the receiver. To be able to steer a beam toward a user, we need to know his/her relative direction to the AP. Recent work has utilized multi-antenna deployments to track a mobile device with high accuracy. For example, ArrayTrack \cite{xiong2013arraytrack}, uses multiple multi-antenna APs to localize a device (smartphone) in office environment, with an error bound of ~20cm. Each AP calculates the angle of arrival of the WiFi signal independently; these estimates are then combined to infer the device's precise location. More recently, the Chronos prototype~\cite{vasisht2016} uses sub-nanosecond estimation of time of flight to perform indoor localization of client devices, using a single AP, to a median accuracy of 65 cm. However, these approaches require active RF transmissions by the mobile device.
Harvesting power from wireless transmissions has also been studied and
usually requires custom-designed hardware for the goal of charging RFID tags
and devices -- with WISP~\cite{sample2008} being a very well known example
that is used to power a variety of sensors. PoWiFi~\cite{talla2015powering}
modified AP firmware to harvest energy from WiFi transmissions without the
use of beamforming. Using beamforming to increase energy harvesting yield
has been studied via simulations by Huang et.
al~\cite{huang2016performance} and Liu et. al~\cite{liu2014multi}. To the
best of our knowledge, \name is the first working prototype.
\subsection{WiFi-based Localization}
More recently, researchers have also investigated \emph{device-free} approaches, which recognize an individual's location or gestures without requiring the user to wear any device. As an example, the WiSee system~\cite{pu2013whole} proposed a technique to recognize hand gestures, based on the doppler shift of a transmitted signal that is captured and analyzed by a separate receiver device. To support operation on a single AP, past work~\cite{bharadia2013full,jain2011practical} has demonstrated the ability to achieve full-duplex RF transmission, which implies that a single WiFi AP's antennas can be used to both transmit and receive (the reflected signal) simultaneously. Using such techniques, recent approaches such as IndoTrack~\cite{li2017} have demonstrated how passive tracking of RF signals reflected by the human body can be used for localization. However, such device-free approaches are still not robust enough to be deployed in arbitrary environments, especially when multiple human occupants are present.
\name requires accurate tracking of a wearable, potentially mobile, device,
to perform accurate beamforming to relieve sufficient RF energy. Prior
work, such as ArrayTrack~\cite{xiong2013arraytrack} and
Chronos~\cite{vasisht2016} have shown how to leverage active client RF
transmissions, coupled with precise AoA computations. to very precisely
locate the client. We use similar methods in \names. Device-free
localisation approaches, such as WiSee~\cite{pu2013whole}, and single AP
methods, such as Bharadia et. al~\cite{bharadia2013full}, Jain et.
al~\cite{jain2011practical}, and IndoTrack~\cite{li2017} were also
considered. But they were not robust enough for our arbitrary deployment
environment with multiple human occupants.
%\subsection{Simultaneous Wireless Information and Power Transfer (SWIPT)}
%\am{Jie--can you add a little text and some references here?}
\ No newline at end of file
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment