Kiran K. Rachuri

SociableSense: exploring the trade-offs of adaptive sampling and computation offloading for social sensing

The interactions and social relations among users in workplaces have been studied by many generations of social psychologists. There is evidence that groups of users that interact more in workplaces are more productive. However, it is still hard for social scientists to capture fine-grained data about phenomena of this kind and to find the right means to facilitate interaction. It is also difficult for users to keep track of their level of sociability with colleagues. While mobile phones offer a fantastic platform for harvesting long term and fine grained data, they also pose challenges: battery power is limited and needs to be traded-off for sensor reading accuracy and data transmission, while energy costs in processing computationally intensive tasks are high. In this paper, we propose SociableSense, a smart phones based platform that captures user behavior in office environments, while providing the users with a quantitative measure of their sociability and that of colleagues. We tackle the technical challenges of building such a tool: the system provides an adaptive sampling mechanism as well as models to decide whether to perform computation of tasks, such as the execution of classification and inference algorithms, locally or remotely. We perform several micro-benchmark tests to fine-tune and evaluate the performance of these mechanisms and we show that the adaptive sampling and computation distribution schemes balance trade-offs among accuracy, energy, latency, and data traffic. Finally, by means of a social psychological study with ten participants for two working weeks, we demonstrate that SociableSense fosters interactions among the participants and helps in enhancing their sociability.

Smartphones for Large-Scale Behavior Change Interventions

Equipped with cutting-edge sensing technology and high-end processors, smartphones can unobtrusively sense human behavior and deliver feedback and behavioral therapy. The authors discuss two applications for behavioral monitoring and change and present UBhave, the first holistic platform for large-scale digital behavior change intervention.

MobileMiner: mining your frequent patterns on your phone

Smartphones can collect considerable context data about the user, ranging from apps used to places visited. Frequent user patterns discovered from longitudinal, multi-modal context data could help personalize and improve overall user experience. Our long term goal is to develop novel middleware and algorithms to efficiently mine user behavior patterns entirely on the phone by utilizing idle processor cycles. Mining patterns on the mobile device provides better privacy guarantees to users, and reduces dependency on cloud connectivity. As an important step in this direction, we develop a novel general-purpose service called MobileMiner that runs on the phone and discovers frequent co-occurrence patterns indicating which context events frequently occur together. Using longitudinal context data collected from 106 users over 1--3 months, we show that MobileMiner efficiently generates patterns using limited phone resources. Further, we find interesting behavior patterns for individual users and across users, ranging from calling patterns to place visitation patterns. Finally, we show how our co-occurrence patterns can be used by developers to improve the phone UI for launching apps or calling contacts.

METIS: Exploring mobile phone sensing offloading for efficiently supporting social sensing applications

Mobile phones play a pivotal role in supporting ubiquitous and unobtrusive sensing of human activities. However, maintaining a highly accurate record of a users behavior throughout the day imposes significant energy demands on the phones battery. In this paper, we present the design, implementation, and evaluation of METIS: an adaptive mobile sensing platform that efficiently supports social sensing applications. The platform implements a novel sensor task distribution scheme that dynamically decides whether to perform sensing on the phone or in the infrastructure, considering the energy consumption, accuracy, and mobility patterns of the user. By comparing the sensing distribution scheme with sensing performed solely on the phone or exclusively on the fixed remote sensors, we show, through benchmarks using real traces, that the opportunistic sensing distribution achieves over 60% and 40% energy savings, respectively. This is confirmed through a real world deployment in an office environment for over a month: we developed a social application over our frameworks, that is able to infer the collaborations and meetings of the users. In this setting the system preserves over 35% more battery life over pure phone sensing.

DSP.Ear: leveraging co-processor support for continuous audio sensing on smartphones

The rapidly growing adoption of sensor-enabled smartphones has greatly fueled the proliferation of applications that use phone sensors to monitor user behavior. A central sensor among these is the microphone which enables, for instance, the detection of valence in speech, or the identification of speakers. Deploying multiple of these applications on a mobile device to continuously monitor the audio environment allows for the acquisition of a diverse range of sound-related contextual inferences. However, the cumulative processing burden critically impacts the phone battery. To address this problem, we propose DSP.Ear -- an integrated sensing system that takes advantage of the latest low-power DSP co-processor technology in commodity mobile devices to enable the continuous and simultaneous operation of multiple established algorithms that perform complex audio inferences. The system extracts emotions from voice, estimates the number of people in a room, identifies the speakers, and detects commonly found ambient sounds, while critically incurring little overhead to the device battery. This is achieved through a series of pipeline optimizations that allow the computation to remain largely on the DSP. Through detailed evaluation of our prototype implementation we show that, by exploiting a smartphones co-processor, DSP.Ear achieves a 3 to 7 times increase in the battery lifetime compared to a solution that uses only the phones main processor. In addition, DSP.Ear is 2 to 3 times more power efficient than a naïve DSP solution without optimizations. We further analyze a large-scale dataset from 1320 Android users to show that in about 80-90% of the daily usage instances DSP.Ear is able to sustain a full day of operation (even in the presence of other smartphone workloads) with a single battery charge.

Sense and sensibility in a pervasive world

The increasing popularity of location based social services such as Facebook Places, Foursquare and Google Latitude, solicits a new trend in fusing social networking with real-world sensing. The availability of a wide range of sensing technologies in our everyday environment presents an opportunity to further enrich social networking systems with fine-grained real-world sensing. However, the introduction of passive sensing into a social networking application disrupts the traditional, user-initiated input to social services, raising both privacy and acceptability concerns. In this work we present an empirical study of the introduction of a sensor-driven social sharing application within the working environment of a research institution. Our study is based on a real deployment of a system that involves location tracking, conversation monitoring, and interaction with physical objects. By utilizing surveys, interviews and experience sampling techniques, we report on our findings regarding privacy and user experience issues, and significant factors that can affect acceptability of such services by the users. Our results suggest that such systems deliver significant value in the form of self reflection and comparison with others, while privacy concerns are raised primarily by the limited control over the way individuals are projected to their peers.

Smartphone sensing offloading for efficiently supporting social sensing applications

Mobile phones play a pivotal role in supporting ubiquitous and unobtrusive sensing of human activities. However, maintaining a highly accurate record of a users behavior throughout the day imposes significant energy demands on the phones battery. In this work, we investigate a new approach that can lead to significant energy savings for mobile applications that require continuous sensing of social activities. This is achieved by opportunistically offloading sensing to sensors embedded in the environment, leveraging sensing that may be available in typical modern buildings (e.g., room occupancy sensors, RFID access control systems). In this article, we present the design, implementation, and evaluation of METIS: an adaptive mobile sensing platform that efficiently supports social sensing applications. The platform implements a novel sensor task distribution scheme that dynamically decides whether to perform sensing on the phone or in the infrastructure, considering the energy consumption, accuracy, and mobility patterns of the user. By comparing the sensing distribution scheme with sensing performed solely on the phone or exclusively on the fixed remote sensors, we show, through benchmarks using real traces, that the opportunistic sensing distribution achieves over 60% and 40% energy savings, respectively. This is confirmed through a real world deployment in an office environment for over a month: we developed a social application over our frameworks, that is able to infer the collaborations and meetings of the users. In this setting the system preserves over 35% more battery life over pure phone sensing.

LEO: scheduling sensor inference algorithms across heterogeneous mobile processors and network resources

Mobile apps that use sensors to monitor user behavior often employ resource heavy inference algorithms that make computational offloading a common practice. However, existing schedulers/offloaders typically emphasize one primary offloading aspect without fully exploring complementary goals (e.g., heterogeneous resource management with only partial visibility into underlying algorithms, or concurrent sensor app execution on a single resource) and as a result, may overlook performance benefits pertinent to sensor processing. We bring together key ideas scattered in existing offloading solutions to build LEO -- a scheduler designed to maximize the performance for the unique workload of continuous and intermittent mobile sensor apps without changing their inference accuracy. LEO makes use of domain specific signal processing knowledge to smartly distribute the sensor processing tasks across the broader range of heterogeneous computational resources of high-end phones (CPU, co-processor, GPU and the cloud). To exploit short-lived, but substantial optimization opportunities, and remain responsive to the needs of near real-time apps such as voice-based natural user interfaces, LEO runs as a service on a low-power co-processor unit (LPU) to perform both frequent and joint schedule optimization for concurrent pipelines. Depending on the workload and network conditions, LEO is between 1.6 and 3 times more energy efficient than conventional cloud offloading with CPU-bound sensor sampling. In addition, even if a general-purpose scheduler is optimized directly to leverage an LPU, we find LEO still uses only a fraction (< 1/7) of the energy overhead for scheduling and is up to 19% more energy efficient for medium to heavy workloads.

Smart phone based systems for social psychological research: challenges and design guidelines

Social psychology research deals with understanding many aspects of human behavior, and this helps not only to gain insights into this complex phenomenon but also to provide useful feedback to the participants. Social psychological research is mainly conducted through self-reports and surveys, however, this methodology is laborious and requires considerable offline analysis. Moreover, self-reports are also found to be biased towards pleasant experiences. Mobile phones represent a perfect platform for conducting social psychological research as they are ubiquitous, unobtrusive, and sensor-rich devices. However, limited battery and computing power, and expensive data plans make it difficult to support various demanding sensing and computation requirements of the social psychological research. In this paper, we describe the specific challenges in building systems based on off-the-shelf mobile phones for conducting social experiments, and provide design guidelines based on our recent works for implementing such systems.

Energy-Accuracy Trade-offs of Sensor Sampling in Smart Phone Based Sensing Systems

A large number of context-inference applications run on off-the-shelf smart phones and infer context from the data acquired by sensing from the sensors embedded in these devices. The use of efficient and effective sampling techniques is of key importance for these applications. Aggressive sampling can ensure a more fine-grained and accurate reconstruction of context information but, at the same time, continuous querying of sensor data might lead to rapid battery depletion. In this chapter, we present a design methodology to evaluate energy-accuracy trade-offs for querying sensor data in continuous sensing mobile systems, and an adaptive sensor sampling methodology that relies on dynamic selection of sampling functions depending on history of context events. We also report on the experimental evaluation of a set of functions that control the rate at which the data are sensed from the accelerometer, Bluetooth, and microphone sensors, and we show that a dynamic adaptation mechanism provides a better energy-accuracy trade-offs compared to simpler function based rate control methods. Furthermore, we show that the suitability of these mechanisms varies for each of the sensors.