SenSys - A Sensor Data Sharing System for Smart Devices using Bluetooth Low Energy Hari Prabhat Gupta, Shivakumar S. Bagi, and Avinash Uttav Samsung R&D Institute India-Bangalore
{harip.gupta,b.shivakumar,a.uttav}@samsung.com
ABSTRACT
or bright) consumes more power and can be harmful to the user’s eyes.
In this work, we present SenSys, a Sensor data sharing System for smart devices. The SenSys opportunistically shares the sensor data of a smart device to other nearby smart devices using Bluetooth Low Energy. The SenSys has been evaluated by well-designed practical experiments to share the sensor data. The prototype results show that the SenSys requires very limited time for sharing the sensor data, reduces the power consumption, and increases the accuracy.
• Sensor data sharing helps to provide the equal power consumption between multiple users. For example, a Geo-fencing image gallery application requires camera and Global Positioning System (GPS) for capturing the images and tagging the location of the users, respectively. By using the data sharing, a smart device uses camera and other nearby smart device communicates GPS coordinates for tagging the location. • The healthcare applications require to share the sensors data (e.g., heart rate, blood pressure, or sudden fall detect) to other devices for monitoring the health of the patients.
Categories and Subject Descriptors C.m [Computer Systems Organization]: MiscellaneousMobile Sensing Systems
• Sharing the sensor data between nearby smart devices can remove the redundant monitoring of the same region by multiple devices (e.g., GPS, humidity, or radiation level) and therefore reduces the power consumption.
General Terms Algorithms, Design, Experimentation, Measurement
Keywords
• The accuracy of some sensors depends on the user environments. For example, magnetometer sensor provides wrong sensor data when a user is close with a metal object. In such scenarios, a user can improve the accuracy by selecting the majority or filtering of sensor data of nearby devices. Another benefit of sensor data sharing is to select the best sensor among the available sensors in nearby smart devices.
Sensing, Resource Sharing, Smart Devices
1.
INTRODUCTION
Smart devices are now used as sensing devices in many applications such as context-awareness, healthcare, home security, and traffic sensing. Some of these applications require the sensor data from other users or share the sensor data to other users. The following points illustrate the requirement of sensor data sharing between nearby smart devices:
The above points illustrate the motivation of the sensor data sharing between nearby smart devices. In this work, we present SenSys, a system for sharing the required sensor data between smart devices. Different from the other existing sensor data sharing systems [1, 2], SenSys works as a plug and play system, i.e., it does not require any user involvement, cloud control, or group formation for sharing the sensor data. In SenSys, a smart device shares the sensor data to other nearby smart devices whenever the sensor data has changed. The rest of this paper is organized as follows: In the next section, we present the overview of the SenSys. Section 3 illustrates the preliminary results. Section 4 concludes the paper.
• Smart devices such as low-cost smartphones, laptops, and televisions have limited sensors. These devices cannot automatically detect the environment changes. For example, old laptops or low-cost smartphones do not have an ambient light sensor and therefore cannot control the display’s brightness. The default display’s brightness setting in different environment (dark Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from
[email protected]. CSAR’15, November 1, 2015, Seoul, South Korea.. c 2015 ACM. ISBN 978-1-4503-3842-4/15/11 ...$15.00.
DOI: http://dx.doi.org/10.1145/2820716.2820728.
2.
SENSYS OVERVIEW
We select Android platform for the implementation of SenSys because it is currently one of the most widely adopted
53
smart device platforms. We use Bluetooth Low Energy (BLE) for sharing the sensor data. We select BLE because it consumes less power compared with other existing techniques (e.g., Classic Bluetooth or Wi-Fi Direct) in smart devices. In SenSys, the sender and receiver of sensor data are known as supplier and consumer, respectively.
Supplier
Consumer
Activity launched
Activity launched
onCreate() onStart()
onCreate() onStart()
Activity in foreground Read sensors Advertise
1 2
Connection request Connection response
3
Discover services List of services
Activity in foreground Require sensor data Scan for device
3.
5
Raw sensor data Std. sensor data
onStop() onDestroy() Activity destroyed
ge chan n Dataificatio not
PRELIMINARY RESULTS
We implement SenSys on smartphones (Samsung Note 4, Samsung Galaxy S5, and Samsung Galaxy S6) running the Android lollipop system for sharing the sensor data of light sensor and GPS coordinates. We assume that the ambient light sensor and GPS of consumer device are not working or switched off. When the standard light sensor data are shared by the supplier, the consumer will be first stored in separate sensor data arrays on the smartphone’s memory. When the length of the data array reaches a threshold, the filtering will be triggered on consumer smartphone. After filtering the sensor data, the consumer will be used for display’s brightness setting. Figure 2 illustrates the standard and filtered sensor data of light sensor. Figure 3 shows the Graphical User Interface (GUI) of the SenSys for sharing the light sensor data, where the supplier and consumer are Samsung Note 4 and Samsung Galaxy S5, respectively.
Set notification Notification success
4
If change in sensor data
quired for sharing the new sensor data. Figure 1 illustrates 1 × 1 connection, i.e., one-supplier and one-consumer. The SenSys also supports M × N connections, where {M, N } ≥ 1. The grater values of M and N provide higher accuracy and reduce the power consumption of SenSys, respectively. As the same type of sensor in different smart devices behaves slightly different, a supplier uses the following step to convert the raw sensor data to a standard sensor data, which is suitable for all kinds of devices. Let M axs and Raws are the maximum value of sensor data and raw sensor data of a sensor s, respectively. The standard sensor data in SenSys is given by (M axs − Raws ) × 100/M axs . After the acquisition of standard sensor data from suppliers, the consumer needs a preprocessing step to remove various kinds of noise. The SenSys uses the Kalman filter for removing such noise from the sensor data as shown in Step 5 of Figure 1.
Std. sensor data Noise Filter Sensor data
onStop() onDestroy() Activity destroyed
14 12
Figure 1: Illustration of the sensor data sharing.
Light sensor data
10
Figure 1 illustrates the sensor data sharing between supplier and consumer devices using BLE on Android platform. Initially, a supplier periodically broadcasts advertisement packets to show its presence to the nearby smart devices (Step 1). Next, the consumer sends a connection request to the supplier. A connection is established between the supplier and consumer (Step 2). The consumer sends discovery request to identify the available services of the supplier. The supplier sends the list of available services (Step 3). The consumer sends notification if the required services are available (Step 4). Steps 1-4 are one time process between supplier and consumer. Step 5 uses to share the sensor data. The SenSys works on push-based technique, i.e., step 5 is automatically repeated whenever the sensor data change. For example, the ambient light sensor manager of the supplier calls step 5 whenever the environment light intensity changes. The main benefit of SenSys is to reduce the power consumption because the consumer does not require any action periodically for monitoring the changes in the sensor data. Similarly, SenSys also reduces the time re-
8 6 4 2 0
Standard sensor data Filtered sensor data 50
100
150 200 Sensor readings
250
300
350
Figure 2: Illustration of standard and filtered sensor data of light sensor.
We measured the power consumption of SenSys using a Monsoon power monitor. We used Samsung Note 4, Samsung Galaxy S5, and Samsung Galaxy S6 as smart devices. We estimate the power consumption for sharing the GPS co-
54
(a) Supplier
Table 2: Illustration of the accuracy of sensor data using SenSys. No. of Suppliers Light Sensor GPS 10 97.53% 98.02% 9 95.31% 96.09% 8 93.84% 94.72% 7 92.78% 93.68% 6 91.99% 92.88% 5 91.38% 92.24% 4 90.89% 91.71% 3 90.48% 91.28% 2 90.15% 90.90% 1 89.86% 90.59%
(b) Consumer
Figure 3: GUI of SenSys for sharing the light sensor. number of suppliers increases the delay of SenSys. This is because the consumer takes time for collecting and filtering the sensor data from multiple suppliers. We conclude from this result that the smart devices require almost equal time for sharing the GPS and light sensor data.
ordinates. We repeat the experiment by change the supplier and consumer smart devices.
3.1
Power Consumption
Table 1 illustrates the reduces of power consumption when M suppliers and N consumers are sharing GPS data, where 1 ≤ {M, N } ≤ 4. From this experiment, we conclude that the SenSys reduces around 32%, 52%, 61%, and 67% power consumption when one, two, three, and four consumes and one supplier are presented, respectively. An interesting observation from this result is that the large number of consumers and suppliers reduces and increases the power consumption, respectively. This is because the large number of consumers switched off the GPS and therefore reduces the power consumption.
Table 3: Illustration of the delay time in milliseconds. No. of Suppliers Light Sensor GPS 5 106 106 4 102 106 3 101 100 2 94 95 1 94 92
Table 1: Illustration of the reduces of power consumption using SenSys. No. of Consumers 1 2 3 4 1-Supplier 32% 52% 61% 67% 2-Suppliers 21% 39% 49% 56% 3-Suppliers 18% 31% 41% 48% 4-Suppliers 14% 26% 35% 42%
3.2
4.
This work presents a system for opportunistically sharing the sensor data between the nearby smart devices using bluetooth low energy. The system evaluated by welldesigned practical experiments to share the sensor data. In the proposed system, a consumer connects with the suppliers and collects the required sensor data using bluetooth low energy. The system uses the Kalman filter for removing the noise from the sensor data. The system works on push-based technique and the consumer does not require any action periodically for monitoring the changes in the sensor data. The prototype results showed that the proposed system requires very limited time for sharing the sensor data, reduces the power consumption, and increases the accuracy. It showed that the system reduces the power consumption up to 67% when four or more consumers are connected. The prototype results also showed that accuracy of the system is achieved up to 98% for sharing the light and GPS sensor data. In the future, we would like to develop sensor data sharing system that can prioritize user chosen sensors.
Accuracy of SenSys
Next, we measure the accuracy of the sensor data using SenSys. We used light and GPS sensors. Table 2 illustrates the accuracy of the sensor data. From this experiment, we conclude that the minimum and maximum accuracy are 89% and 98%, respectively. The result shows that the accuracy of the GPS is higher than the light sensor. This is because the GPS value is almost same within the BLE range. We conclude from this result that the large number of suppliers increases the accuracy of the proposed SenSys.
3.3
CONCLUSION
5.
Delay
REFERENCES
[1] S. Hemminki and et al. Cosense: a collaborative sensing platform for mobile devices. In Proc. of SenSys, pages 34:1–34:2, 2013. [2] E. Miluzzo and et al. Darwin phones: the evolution of sensing and inference on mobile phones. In Proc. of MobiSys 2010, pages 5–20, 2010.
Finally, We measured the required time for sharing the sensor data between the smart devices by using the Android dumpstate logs. Table 3 illustrates the sensor data delay time in milliseconds. It illustrates that maximum delay and the average sensor data sharing time are 106 and 100 milliseconds, respectively. Table 3 also shows that the large
55
