Battery-Free Cellphone

Abstract

We present the first battery-free cellphone design that consumes only a few micro-watts of power. Our design can sense speech, actuate the earphones, and switch between uplink and downlink communications, all in real time. Our system optimizes transmission and reception of speech while simultaneously harvesting power which enables the battery-free cellphone to operate continuously. The battery-free device prototype is built using commercial-off-the-shelf components on a printed circuit board. It can operate on power that is harvested from RF signals transmitted by a basestation 31 feet (9.4 m) away. Further, using power harvested from ambient light with tiny photodiodes, we show that our device can communicate with a basestation that is 50 feet (15.2 m) away. Finally, we perform the first Skype call using a battery-free phone over a cellular network, via our custom bridged basestation. This we believe is a major leap in the capability of battery-free devices and a step towards a fully functional battery-free cellphone. CCS Concepts: •Computer systems organization →Embedded and cyber-physical systems; •Human-centered computing →Ubiquitous and mobile computing;

Battery-free Visible Light Sensing

“…Processing overhead. Prior attempts [10,15] have investigated the processing overhead on backscatter sensors. Zhang et al optimise the computational block, and demonstrate that processing is more energy expensive than backscatter transmissions.…”

“…Zhang et al optimise the computational block, and demonstrate that processing is more energy expensive than backscatter transmissions. Talla et al [10], in a concurrently released work, design a battery-free cellphone. Their design is similar to Scatterlight in terms of eliminating computational blocks.…”

“…Existing VLS systems also consume significant energy to process and convert analog signals to the digital domain due to the use of energy expensive ADCs [7,8]. The overhead is substantial for sensing systems [10] where high frequency ADC sampling operations are required to detect shadows or other sensing events. Digitising at µWs of power.…”

“…Processing sensor values is energy expensive due to computational blocks [10,15]. Existing VLS systems leverage platforms such as an Arduino to preprocess light samples, and transfer these to an end-device such as a workstation [7,8].…”

“…However, the key challenge with such a design is that backscatter signals are at the same frequency as the carrier signal which causes severe selfinterference [10]. Further, supporting multiple VLMs becomes challenging as the tags backscatter at the same frequency.…”