Andreas Soleiman

Currently, I am a full time Data Scientist at H&M Group laboratory, an innovative arm within the organization. I am involved in the design and development of new tools for deployment in production. The work includes researching latest industry practices and machine learning based tools, studying internal and external data sets, looking for patterns and writing the code for prototyping new ideas, generate predictions, evaluate the test sets, perform advanced analytics, etc.

Between Jan 2020 - Jan 2021, I received admittance as a PhD student at MIT, from which I took a leave of absence. During this time, I participated in research internships in Germany (MPI-SWS) and UK (Oxford/Cambridge) to further work on battery-free sensing for mobile sensing and on device ML.

Between June 2017 to Dec 2019, I have been a researcher at Uppsala University where I worked on low-power mobile networking and visible light sensing systems, advised by Dr. Ambuj Varshney. During this period, I also received a Masters' degree in Engineering Physics, which includes a mix of applied physics, electrical engineering and computer science.

Email  /  CV  /  LinkedIn

• April 2022: Data Scientist at H&M Group


• December 2021: Software Engineer at Zenseact


• January 2021: Fully employed as a Data Scientist at Electrolux


• January 2021: Wrote a blog post about my personal journey through the pandemic as an incoming student at MIT!


• June 2020: Accepted for a summer internship position at the University of Cambridge under the supervision of Prof. Nicholas Lane


• March 2020: Offered a PhD position at Massachusetts Insitute of Technology (MIT)!


• January 2020: Accepted for an internship position at the Max Planck Institute for Software Systems (MPI-SWS) under the supervision of Prof. Peter Druschel


• December 2019: Invited to the University of Oxford to give a talk about my work in battery-free sensing


• November 2019: My supervisor Ambuj won the prestigious ABB Research Award for his tireless work on sustainable networked embedded systems. I am very proud to have worked closely with him for the past 2.5 years at Uppsala!


• August 2019: Our recent work on the design of a low power communication mechanism for self-sustaining sensors is accepted as a full paper at ACM MobiCom 2019


• June 2019: Our work on battery-free utensils has been featured by the largest electronic newspaper in Sweden


• April 2019: Poster and demonstration was accepted at ACM MobiSys 2019


• April 2019: Invited as PC member of the ACM S3 2019 workshop, held in conjunction with ACM MobiCom 2019

• Selected to attend the Rising Stars Forum of ACM MobiSys 2019


• Selected for the Cornell, Maryland, Max Planck Pre-Doctoral Research School 2018


• Best Demo Award, ACM WiSec 2018


• Best Paper Award, ACM VLCS 2017, held in conjunction with ACM MobiCom 2017


• Winner of the Student Research Competition at ACM MobiCom 2017

• External Reviewer: ACM Interactive, Mobile, Wearable and Ubiquitous Technologies (IMWUT): 2019, 2020


• PC Member: ACM S3 2019 workshop, held in conjunction with ACM MobiCom 2019

Due to extremely low power consumption, backscatter has become the transmission mechanism of choice for battery-free devices that operate on harvested energy. However, a limitation of recent backscatter systems is that the communication range scales with the strength of the ambient carrier signal (ACS). This means that to achieve a long range, a backscatter tag needs to reflect a strong ACS, which in practice means that it needs to be close to an ACS emitter. We present TunnelScatter, a mechanism that overcomes this limitation. TunnelScatter uses a tunnel diode-based radio frequency oscillator to enable transmissions when there is no ACS, and the same oscillator as a reflection amplifier to support backscatter transmissions when the ACS is weak. Our results show that even without an ACS, TunnelScatter is able to transmit through several walls covering a distance of 18 meters while consuming a peak biasing power of 57 microwatts. Based on TunnelScatter, we design battery-free sensor tags, called TunnelTags, that can sense physical phenomena and transmit them using the TunnelScatter mechanism.

In this early-stage work, we propose various solutions to enable the next generation of wireless sensors. Our vision is to introduce battery-free wireless sensors that can be deployed ubiquitously. Such sensors would have the ability to both infer the physical environment, and communicate the sensed information wirelessly. In particular, we explore the emerging research directions of ambient and analog RF backscatter for communication, and visible light for sensing. We combine these concepts with energy harvesting to achieve self-powered operation. Furthermore, we introduce novel mechanisms that eliminate sensor-local computational blocks, and instead couple sensors directly to ultra-low power communication modules to transmit sensor information. Our initial results show that we are able to achieve operation of both sensing and communication at a few microwatts of power. Moreover, we can maintain a sufficiently high sensing resolution to enable novel battery-free applications such as hand gesture sensing and intrusion detection.

Backscatter communication enables wireless transmissions at orders of magnitude lower power consumption when compared to conventional radio transceivers. This introduces novel opportunities for battery-free and ubiquitous sensing. We take advantage of backscatter communication to enable networked utensils. We imagine a scenario where such utensils can provide essential information about the state of the food or the beverage; for instance, the temperature or the quality of food contained in the utensils. We propose flex sensors, to achieve this capability, by augmenting utensils with flexible and inexpensive battery-free sensors that can communicate wirelessly. We demonstrate our efforts by designing a smart cup that tracks the temperature of beverage.

Light as a medium for sensing and communication enables new scenarios, such as controlling devices with gestures, or communication for Internet of Things~(IoT) devices. However, a limitation of existing systems is that they often sense only a narrow part of the light spectrum. We argue that the ability to sense a broad light spectrum significantly enhances ability of such systems expanding possible application scenarios. We demonstrate our work in progress to develop the concept of polymorphic light sensing~(PLS). PLS sensor morphs itself according to applications requirements, to track desired parts of the light spectrum (colours, infrared, and ultraviolet light). We couple the PLS sensor with ultra-low power backscatter mechanism, and demonstrate this enables us to sense and communicate the broad spectrum, while operating battery-free.

Radio Tomographic Imaging (RTI) enables novel radio frequency (RF) sensing applications such as intrusion detection systems by observing variations in radio links caused by human actions. RTI applications are, however, severely limited by the requirement to retrofit existing infrastructure with energy-expensive sensors. In this demonstration, we present our ongoing efforts to develop the first battery-free RTI system that operates on minuscule amounts of energy harvested from the ambient environment. Our system eliminates the energy-expensive components employed on state-of-the-art RTI systems achieving two orders of magnitude lower power consumption. Battery-free operation enables a sustainable deployment, as RTI sensors could be deployed for long periods of time with little maintenance effort. Our demonstration showcases an intrusion detection scenario enabled by our system.

We present the first visible light sensing system that can sense and communicate shadow events while only consuming tens of microwatts of power. Our system requires no modification to the existing lighting infrastructure and can use unmodulated ambient light as a sensing medium. We achieve this by designing a sensing mechanism that utilizes solar cells, and an ultra-low power backscatter based transmission mechanism we call Scatterlight, which can communicate sensor readings without the use of any energy-expensive computational block. Our results demonstrate the ability to sense and communicate various hand gestures at peak power consumption of tens of microwatts at the sensor, which represents orders of magnitude improvement over the state-of-the-art.

We present the design of the first Visible Light Sensing (VLS) system that consumes only tens of microwatts of power to sense and communicate. Unlike most existing VLS systems, we require no modification to the existing light infrastructure since we useunmodulated light as a sensing medium. We achieve this by designing a novel mechanism that uses solar cells to achieve a sub-microwatt power consumption for sensing. Further, we devise an ultra-low power transmission mechanism that backscatters sensor readings and avoids the processing and computational overhead of existing sensor systems. Our initial results show the ability to detect and transmit hand gestures or presence of people up to distances of 330 meter, at a peak power of 20 microwatts. Further, we demonstrate that our system can operate in diverse light conditions (100 lx to 80 klx) where existing VLS designs fail due to saturation of the transimpedance amplifier (TIA).