Toward Polymorphic Internet of Things Receivers Through Real-Time Waveform-Level Deep Learning

Wireless systems such as the Internet of Things (IoT) are changing the way we interact with the cyber and the physical world. As IoT systems become more and more pervasive, it is imperative to design wireless protocols that can effectively and efficiently support IoT devices and operations. On the other hand, today's IoT wireless systems are based on inflexible designs, which makes them inefficient and prone to a variety of wireless attacks. In this paper, we introduce the new notion of a deep learning-based polymorphic IoT receiver, able to reconfigure its waveform demodulation strategy itself in real time, based on the inferred waveform parameters. Our key innovation is the introduction of a novel embedded deep learning architecture that enables the solution of waveform inference problems, which is then integrated into a generalized hardware/software architecture with radio components and signal processing. Our polymorphic wireless receiver is prototyped on a custom-made software-defined radio platform. We show through extensive over-the-air experiments that the system achieves throughput within 87% of a perfect-knowledge Oracle system, thus demonstrating for the first time that polymorphic receivers are feasible.

Efficient Wideband Spectrum Sensing Using Mems Acoustic Resonators

The ever-increasing demand for wireless applications has resulted in an unprecedented radio frequency (RF) spectrum shortage. Ironically, at the same time, actual utilization of the spectrum is sparse in practice [1]. To exploit previously underutilized frequency bands to accommodate new unlicensed applications and achieve highly efficient usage of the spectrum, the Federal Communications Committee (FCC) has repurposed many frequency bands for dynamic spectrum sharing. This includes the 6 GHz band to be shared between Wi-Fi 6 and the incumbent users [2] as well as the 3.5 GHz Citizens Broadband Radio Service (CBRS) band [3].

When Tiny Goes Big

I believe that the Internet of Tiny Things (IoTT) will be the next big driver of the computing and semiconductor industry - imagine applications such as smart city, home sensors, wearables, implantables, single-use electronic tags for pharmaceuticals and produce, and more. Trillions of tiny devices may be needed every year to enable these applications, while meeting extreme requirements in terms of cost (sometimes only a few cents), power (often self-powered), and trust (often physically accessible and producing sensitive data). Our research over last few years has been focused on enabling an internet of these tiny things by addressing the unique cost, power, and trust challenges of these devices.

Expanding the Horizons of Wireless Sensing

All interactions of objects, humans, and machines with the physical world are via contact forces. For instance, objects placed on a table exert their gravitational forces, and the contact interactions via our hands/feet are guided by the sense of contact force felt by our skin. Thus, the ability to sense the contact forces can allow us to measure all these ubiquitous interactions, enabling a myriad of applications. Furthermore, force sensors are a critical requirement for safer surgeries, which require measuring complex contact forces experienced as a surgical instrument interacts with the surrounding tissues during the surgical procedure. However, with currently available discrete point-force sensors, which require a battery to sense the forces and communicate the readings wirelessly, these ubiquitous sensing and surgical sensing applications are not practical. This motivates the development of new force sensors that can sense, and communicate wirelessly without consuming significant power to enable a battery-free design. In this magazine article, we present WiForce, a low-power wireless force sensor utilizing a joint sensing-communication paradigm. That is, instead of having separate sensing and communication blocks, WiForce directly transduces the force measurements onto variations in wireless signals reflecting WiForce from the sensor. This novel trans-duction mechanism also allows WiForce to generalize easily to a length continuum, where we can detect as well as localize forces acting on the continuum. We fabricate and test our sensor prototype in different scenarios, including testing beneath a tissue phantom, and obtain sub-N sensing and sub-mm localizing accuracies (0.34 N and 0.6 mm, respectively).

A Systems Approach to Achieve Equity in Healthcare Research

Healthcare in the United States is inequitable. The consequence of inequity is that the burden of serious chronic disease, such as diabetes, falls disproportionately on populations experiencing health disparities, predominantly Black, Indigenous, and people of color. [1] The reasons for the inequity include the negative impact of the social determinants of health of individuals and families from these communities, being underrepresented as participants in clinical research, having limited access to technologies that support self-care, and a lack of researchers and clinicians from these same populations. [2] To achieve equity and fairness, there is a need for a paradigm shift in healthcare research and innovation based on improving access, trust, and self-efficacy [3] to convert new knowledge into positive health outcomes.

AuraRing

Wearable computing platforms, such as smartwatches and head-mounted mixed reality displays, demand new input devices for high-fidelity interaction. We present AuraRing, a wearable magnetic tracking system designed for tracking fine-grained finger movement. The hardware consists of a ring with an embedded electromagnetic transmitter coil and a wristband with multiple sensor coils. By measuring the magnetic fields at different points around the wrist, AuraRing estimates the five degree-of-freedom pose of the ring. AuraRing is trained only on simulated data and requires no runtime supervised training, ensuring user and session independence. It has a resolution of 0.1 mm and a dynamic accuracy of 4.4 mm, as measured through a user evaluation with optical ground truth. The ring is completely self-contained and consumes just 2.3 mW of power.

An Introduction to the Federated Learning Standard

With the growing concern on data privacy and security, it is undesirable to collect data from all users to perform machine learning tasks. Federated learning, a decentralized learning framework, was proposed to construct a shared prediction model while keeping owners' data on their own devices. This paper presents an introduction to the emerging federated learning standard and discusses its various aspects, including i) an overview of federated learning, ii) types of federated learning, iii) major concerns and the performance evaluation criteria of federated learning, and iv) associated regulatory requirements. The purpose of this paper is to provide an understanding of the standard and facilitate its usage in model building across organizations while meeting privacy and security concerns.

Exploring LoRa for Sensing

Wireless sensing received a great amount of attention in recent years and various wireless technologies have been exploited for sensing, including WiFi [1], RFID [2], ultrasound [3], 60 GHz mmWave [4] and visible light [5]. The key advantage of wireless sensing over traditional sensing is that the target does not need to be equipped with any sensor(s) and the wireless signal itself is being used for sensing. Exciting new applications have been enabled, such as passive localization [6] and contactless human activity sensing [7]. While promising in many aspects, one key limitation of current wireless sensing techniques is the very small sensing range. This is because while both direct path and reflection path signals are used for communication, only the weak target-reflection signals can be used for sensing. Take Wi-Fi as an example: the communication range can reach 20 to 50 meters indoors but its sensing range is merely 4 to 8 meters. This small range further limits the through-wall sensing capability of Wi-Fi. On the other hand, many applications do require long-range and through-wall sensing capability. In a fire rescue scenario, the sensing device cannot be placed close to the building, and the long-range through-wall sensing capabilities are critical for detecting people deep inside the building. Table I summarizes the sensing range of existing wireless technologies. We can see that long-range through-wall sensing is still missing with wireless sensing.

Can You See It?

Today's smartphones and wearable devices come equipped with an array of inertial sensors, along with IMU-based Human Activity Recognition models to monitor everyday activities. However, such models rely on large amounts of annotated training data, which require considerable time and effort for collection. One has to recruit human subjects, define clear protocols for the subjects to follow, and manually annotate the collected data, along with the administrative work that goes into organizing such a recording.

FluSense

Influenza is a highly contagious respiratory infection that leads to regular seasonal epidemics. It is a major contributor to morbidity and mortality, and in the United States, since 2010, it has infected between 9.2 million and 60.8 million people and has caused between 12,000 and 56,000 deaths. [2]. Moreover, the economic impact [1] of influenza is estimated to be 47 billion to 150 billion dollars per year in the USA alone.