Recent research has demonstrated the huge potential of WiFi for contactless sensing of human activities. Unfortunately, such sensing is highly sensitive to the relative orientation between the user and the WiFi receivers. To overcome this problem, existing solutions deploy multiple WiFi receivers at precise positions to capture orientation-independent view of the human activity. Orientation-independent single-receiver WiFi sensing is still considered an open problem. In this article, we propose a deep neural network architecture that uses radar data during training to learn high-precision Doppler features of human activities from the noisy channel states observed by a single WiFi receiver. Once trained with radars, the network can be used to detect human activities at any arbitrary orientations based only on WiFi signals. Using extensive experiments with millimeter-wave radars, we demonstrate that the proposed approach, called WiFi2Radar in this article, significantly outperforms state-of-the-art for detecting human activities in untrained orientations using only a single WiFi receiver. Our results show that WiFi2Radar can detect orientation-independent human activities with up to 91% accuracy, which outperforms the state of the art by 19%.

Automobile manufacturers are building cars with passive keyless entry support – the ability for users to unlock and start their cars while having the car keys "in their pockets". In this regard, car manufacturers are currently installing Ultra-Wideband (UWB) radios in cars owing to their robustness to relay attacks and their precise ranging estimates. In this paper, we explore the possibility of using these UWB radios (that exist for keyless entry) as an orthogonal sensing modality. To test this hypothesis, we build CaraoKey, the first system that exploits the UWB keyless infrastructure as a "sensor". More specifically, CaraoKey leverages the UWB radios to infer the state of a car (empty, door, window or trunk open, person inside the car, etc.). It does so by building a multipath profile based on the Channel Impulse Response (CIR) that is computed by each UWB node. We design and implement a 14-node system (a superset of all possible node locations) and evaluate it at 7 different indoor and outdoor locations while capturing a wide range of factors such as cars, people, walls on the sides, etc. Our results indicate that CaraoKey can distinguish 8 different car states with 98% accuracy using 8 nodes, and nearly 94% accuracy using just 4 nodes.

Many homes today are logically or physically "zoned" based on properties such as HVACs, activities, or physical layouts. Accurately sensing the occupancy of these zones can yield energy savings, aid in automatic heating and lighting control, energy disaggregation, etc. Existing systems that attempt to sense occupancy are power consuming, non real-time, pet unfriendly and/or sensitive to ambient heat, light and air flow. In this paper, we address these by building Doorpler, a time, space and power-aware radar-based system that detects occupancy at zone transition points by sensing crossings and their direction. It detects a crossing via the Doppler Principle, and infers the direction of crossing by measuring the angle-of-arrival of the human reflection. We evaluate Doorpler by conducting a scripted study and two in-situ studies for 200 hours, collecting over 1600 doorway crossings. We obtain a precision, recall and direction accuracy of over 99% in the scripted studies, and over 95% in the in-situ studies. Our results estimate that Doorpler can fall in the energy-harvestable range of indoor environments with an average power consumption of 6.1mW. With an execution time of 13.8ms, Doorpler has the potential to enable several real-time smart home applications like smart-lighting and HVAC control.

Decimeter-level localization has become a reality, in part due to the ability to eliminate the effects of multipath interference. In this paper, we demonstrate the ability to use multipath reflections to enhance localization rather than throwing them away. We present Multipath Triangulation, a new localization technique that uses multipath reflections to localize a target device with a single receiver that does not require any form of coordination with any other devices. In this paper, we leverage multipath triangulation to build the first decimeter-level WiFi localization system, called MonoLoco, that requires only a single access point (AP) and a single channel, and does not impose any overhead, data sharing, or coordination protocols beyond standard WiFi communication. As a bonus, it also determines the orientation of the target relative to the AP. We implemented MonoLoco using Intel 5300 commodity WiFi cards and deploy it in four environments with different multipath propagation. Results indicate median localization error of 0.5m and median orientation error of 6.6 degrees, which are comparable to the best performing prior systems, all of which require multiple APs and/or multiple frequency channels. High accuracy can be achieved with only a handful of packets.

Super-resolution subspace methods are popular in estimating multipath parameters such as angle of arrival and time of flight. However, they require decorrelation techniques to resolve coherent multipath components. The conventional decorrelation techniques reduce the effective aperture of the MIMO array, thus reducing the resolution and number of resolved paths. In this paper, we introduce MIMO smoothing as a new technique to bring decorrelation effect by leveraging the spacial and frequential diversity in MIMO transmitters and receivers. Via extensive experiments on WiFi links, we show that MIMO smoothing can increase the accuracy of multipath resolution.

A large amount of energy could be saved by detecting home occupancy and automatically controlling the lights, HVAC, water heating, and other mechanical systems. Existing systems rely on motion information, which usually fail to detect occupied rooms with stationary people. In this project, we study the possibility of converting commodity WiFi access points to occupancy sensors by exploiting multipath reflections as individual spatial sensors. The proposed method measures fine-grained distortions caused by human body on phase and amplitude of WiFi signals. Our initial results suggest that formulating WiFi parameters into angle of arrival provides a more sensitive metric to measure occupancy.

Doorway tracking systems track people’s room location by instrumenting the doorways rather than instrumenting the rooms themselves—resulting in fewer sensors and less monitoring while still providing location information on occupants. In this article, we explore what is required to make doorway tracking a practical solution. We break a doorway tracking system into multiple independent design components, including both sensor and algorithmic design. Informed by this design, we construct a doorway tracking system and analyze how different combinations of these design components affect tracking accuracy. We perform a six-day in situ study in a ten-room house with two volunteers to analyze how these design components respond to the natural types and frequencies of errors in a real-world setting. To reflect the needs of different application classes, we analyze these design components using three different evaluation metrics: room accuracy, duration accuracy, and transition accuracy. Results indicate that doorway tracking can achieve 99.5% room accuracy on average in controlled settings and 96% room accuracy in in situ settings. This is contrasted against the 76% in situ setting room accuracy of Doorjamb, a doorway tracking system whose design implements only a limited number of components in our proposed doorway tracking system design space. We describe the differences between the data in the in situ and controlled settings, and provide guidelines about how to design a doorway tracking system for a given application’s accuracy requirements.

In this paper, we consider a variant of the multi-target tracking problem in which the tracking region is divided into zones and targets can only be monitored as they transition between these zones. We call this the transition tracking problem. The key challenge in Transition Tracking is to estimate the number of targets in the tracking region without being able to sense all targets simultaneously. In this paper, we propose an approach to the Transition Tracking problem called TransTrack. Unlike most other tracking algorithms that maximize the likelihood of the sensor data, TransTrack applies penalty functions to find the minimum number of targets that can explain the sensor data. These penalties allow tracks with larger numbers of targets only ifthey have sufficiently fewer errors than other, alternative tracks. To evaluate this approach, we apply TransTrack to a data set containing 3275 transitions between rooms in a home. We observe an average room tracking accuracy of up to 94.5%.

With ever expanding datasets, efficient data management in grids becomes important. This paper describes Cabinet which employs two techniques for efficiently managing data in grids-a caching system and a new file staging approach called coordinated staging. The caching system is designed based on the characteristics of grid applications. Coordinated staging is based on the BitTorrent Protocol model and is specifically designed for High Throughput Computing (HTC) applications, a common use-case for grids. In coordinated staging, each site that is assigned to execute an individual job of the HTC application treats other execution sites as potential replica-stores. In our evaluation, we show that coordinated staging lowered the download time of a file by 3.85x, and increased the throughput of the download by 2.86x over the conventional approach of file transfer from a single source.