You are on page 1of 5

2020 International Russian Automation Conference (RusAutoCon)

The Use of Butterworth Filter to Compensate for


Noise in Signals from Bluetooth Low Energy
Beacons in Autonomous Navigation Systems
A. V. Astafiev A. L. Zhiznyakov A. A. Demidov
Department of Physics and Applied Department of Software Engineering Department of Physics and Applied
Mathematics Vladimir State University Mathematics
Vladimir State University Vladimir, Russia Vladimir State University
Vladimir, Russia Lvovich1975@mail.ru Vladimir, Russia
Alexandr.Astafiev@mail.ru AADemidov@list.ru

Abstract—Currently, one of the most popular areas of The first mention of the use of low-energy technologies
scientific research is the Internet of things. This work is aimed at according to information from the scientific electronic library
studying the operation of low-energy beacons operating on the eLibrary [2] dates back to 2015 [3].
basis of Bluetooth Low Energy technology. In view of the fact that
these beacons operate in a very noisy frequency range, it is The first mention of positioning on the basis of low-energy
necessary to conduct additional studies and propose approaches to beacons in foreign scientific literature, according to Springer, is
reducing the signal noise. The paper proposes the use of a also dated 2015. In their article [4], the authors conducted
Butterworth filter to filter the resulting multidimensional series of studies that showed the effectiveness of technology for remote
data on the signal level from the beacon. A mathematical control of gates in a “smart home”. The authors of Schwarz D.,
description of the algorithm is given, signal levels are measured in Schwarz M., Stückler J., Behnke S. in their article [5] attempted
laboratory conditions. Experimental studies of the operation of the to develop the application “Cosero, Find My Keys!” to search
filtering algorithm on the obtained data sets were carried out. for keys by a robot on a limited, pre-marked area.
Keywords—BLE, beacon, indoor positioning, Butterworth filter, In 2019, the number of articles on the use of low-energy
trilaceration, movement control technologies for positioning and navigation (including indoor)
according to Springer publisher was 402 (Scopus, Web of
I. INTRODUCTION Science), according to eLibrary - 138.
According to the program “Development of the Digital Despite the large volume of research, most of them are of a
Economy in Russia until 2035”, one of the main technologies point and experimental nature in order to study the possibility of
that determine the transition to the digital economy is using the technology, and not the theoretical development of
identification technology [1], the development of which methods and algorithms to improve the quality of the results.
methods the project aims to develop. In particular, the Based on this, the paper proposes a consideration of approaches
development of AutoNet, MariNet and EnergyNet markets to filtering signals from Bluetooth Low Energy beacons to
requires the development of positioning systems and compensate for noise.
autonomous navigation [1], such as RTLS (from the English.
Real-time Locating Systems - real-time positioning system). The aim of the work is to conduct a comparative analysis and
The development of theoretical foundations and methods for the the effect of applying digital signal filters to compensate for
development of intelligent autonomous navigation subsystems noise in signals from Bluetooth Low Energy beacons using the
for building RTLS systems has an important role in the Butterworth filter as an example. To achieve the task it is
implementation of many promising projects, such as “smart necessary to solve the following tasks:
manufacturing”, “smart cities”, etc. 1. To analyze the domestic and foreign scientific and
The research in the direction of positioning and navigation technical base on the research topic.
based on low-energy beacons has shown high efficiency, as a 2. Conduct a comparative analysis of digital signal filters.
result of which the "giants" of the industry such as Google and
Apple began to conduct their own research in this area. The 3. Implement a Butterworth filter to compensate for noise in
result of Google’s research was the invention of the EddyStone signals from Bluetooth Low Energy beacons.
lighthouse data processing format - URL transfer (2015). Radius
4. Carry out experimental studies of the developed algorithm
Networks has created the AltBeacon format - the transmission
and compare with similar filters.
of a unique identifier (2014). Apple has developed its iBeacon
data transfer format for transmitting a unique identifier with a II. SUBJECT OVERVIEW
grouping in two groups (2013).
The latest work on organizing navigation based on Bluetooth
Low Energy low-energy beacons can include the following.

978-1-7281-6130-3/20/$31.00 ©2020 IEEE 1117

Authorized licensed use limited to: University of Durham. Downloaded on November 01,2020 at 14:32:42 UTC from IEEE Xplore. Restrictions apply.
2020 International Russian Automation Conference (RusAutoCon)
Authors Cheng Zhou, Jiazheng Yuan, Hongzhe Liu, Jing 6. Phase filters (FF) - have, in the ideal case, a constant
Qiu in their work [6] consider the use of Bluetooth technology transmission coefficient at all frequencies and is designed to
4.0 for indoor positioning using a Kalman filter to suppress noise change the phase of the input signals (in particular for the time
and the influence of physical barriers on signal level. As a delay of the signals).
positioning algorithm, multilateration based on the readings of
In turn, for processing digital signals can be used:
four beacons is used. Experimental studies were carried out at
the museum. The approach developed by the authors makes it 1. Kalman filter.
possible to accurately implement positioning indoors, however,
a feature of the algorithm is that in most cases the fourth point is 2. Simplified Kalman filter.
very far from the signal receiver, which reduces the accuracy of 3. Butterworth filter.
positioning. Researching a team from the fifth point gives an
even bigger mistake. 4. Three-compartment median filter.
The authors of Hee-Yong Kang, Ji-na Lee, Yoonkyu Kang, 5. Five-compartment median filter.
Jong-Bae Kim, Hyung-Woo Park, Myung-Jin Bae in [7] use 6. Seven-section median filter.
Bluetooth Low Energy technology, distance measurement by
reading the RSSI signal level, pre-processing by Kalman filter 7. Neural networks.
and coordinate prediction (Pedestrian dead reckoning (PDR)). One of the most common approaches is to use a simplified
The studies were conducted in open space - in the stadium. The Kalman filter for a digital signal. The advantage of this filter is
combination of RSSI methods and numbering of coordinates that its use requires a small amount of input data, namely the
allowed the authors to achieve better accuracy, but it did not value of the previous filtered reading. However, this approach is
exceed 2.2 meters. If this development is used indoors, this not recognized as effective, because does not take into account
parameter will be even lower. the specifics of the signal - its mathematical model. In the project
The authors of Haifeng Cong, Liangbo Xie, Mu Zhou in their under consideration, the simplified Kalman filter shows high
work [8] fingerprint method of Bluetooth signals for positioning. accuracy in the absence of movement of the controlled object.
To improve positioning accuracy, a method is proposed for When moving, the simplified Kalman filter “eats” part of the
automatic updating of the fingerprint database of radio signals useful signal.
in view of changes in the external environment that affect signal In the full Kalman filter, a specific mathematical model of
propagation. the process in question is used for forecasting. That is, in the
Authors Patricio Lara-Alvarez, Janio Jadán-Guerrero, César positioning problem under consideration, we must know in
Guevara-Maldonado, Sandra Sanchez-Gordon, Tania Calle- advance the direction and speed of the controlled object, which
Jimenez, Luis Salvador-Ullauri, Patricia Acosta-Vargas, is difficult. The only source of such data is inertial navigation,
Wilmer Illescas Espinoza in May [9] use Bluetooth Low to build which itself requires the use of filtering methods.
a positioning system for people with disabilities. For The Butterworth filter is one of the common electronic filters
implementation, the ready-made proximi.io library is used. in the theory of digital signal processing. The main difference
The authors of Xuyu Wang, Zhitao Yu, and Shiwen Mao in from the Kalman filter is that for filtering it is necessary to
[10] propose, in addition to a network of beacons, to build maps transmit the entire time series of the signal for passage in both
of the illumination and magnetic fields of a controlled room to directions.
aggregate the information received.
III. BUTTERWORTH FILTER ALGORITHM IMPLEMENTATION
Filters are frequency-selective devices that transmit or delay
In order to compensate for emissions and interference in the
signals lying in specific frequency bands.
signals received from BLE beacons, the study proposes to use
Filters can be classified by their frequency characteristics: the Butterworth filter. During the work, a software
implementation of this filter was made. The operation algorithm
1. Low-pass filters (low-pass filters) - they pass all vibrations can be described by the following sequence of actions:
with frequencies not higher than a certain cut-off frequency and
a constant component. 1. The input of the algorithm receives a time series of data
obtained from a specific BLE beacon (IS - Input Signal), the
2. High-pass filters (low-pass filters) - allow all vibrations to interval between measurements (deltaTimeInSec), and the
pass no lower than a certain cut-off frequency. cutoff frequency (cutOff). An example of such a series of data
3. Band-pass filters (PF) - allow fluctuations in a certain is shown in Figure 1.
frequency band, which is determined by a certain level of
frequency response.
4. Band-suppression filters (PPF) - delay oscillations in a
certain frequency band, which is determined by a certain level
of frequency response.
5. Notch filters (RF) - a type of PPF having a narrow delay
band and also called a filter plug.
example of time series of data obtained from a specific BLE beacon

1118

Authorized licensed use limited to: University of Durham. Downloaded on November 01,2020 at 14:32:42 UTC from IEEE Xplore. Restrictions apply.
2020 International Russian Automation Conference (RusAutoCon)
2. The array received at the input is supplemented by four
service elements: 2 on each side.
3. The variables Wc, k1, k2, k3, a, b, c, d, e of the algorithm
are initialized according to the following formulas:

𝑊𝑊𝑊𝑊 = tan(𝑐𝑐𝑐𝑐𝑡𝑡𝑂𝑂𝑂𝑂𝑂𝑂 ∗ 𝜋𝜋 ∗ 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑),

𝑘𝑘1 = √2 ∗ 𝑊𝑊𝑊𝑊,

𝑘𝑘2 = 𝑊𝑊𝑊𝑊 2 ,
𝑘𝑘2
𝑎𝑎 = ,
1+𝑘𝑘1+𝑘𝑘2

𝑏𝑏 = 2 ∗ 𝑎𝑎,

𝑐𝑐 = 𝑎𝑎,
𝑏𝑏
𝑘𝑘3 = ,
𝑘𝑘2

𝑑𝑑 = −2 ∗ 𝑎𝑎 + 𝑘𝑘3,

𝑒𝑒 = 1 − 2 ∗ 𝑎𝑎 − 𝑘𝑘3.

4. The service elements of the temporary direct pass array


(DFS - Direct Filtered Signal) are initialized in such a way that:

DFS [0] = DFS [1] = IS [0],

DFS [length + 2] = DFS [length + 1] = IS [length].

5. Next, each element of the DFS array starting from 3


elements and up to length is calculated by the formula:

DFS [i] = a * IS [i] + b * IS [i-1] + d * DFS [i-1] + c * IS [i-1]


+ e * DFS [i-2].

6. The service elements of the temporary backward pass


array (FFS - Forward Filtered Signal) are initialized in such a
way that:

FFS [length] = DFS [length + 2],

FFS [length + 1] = DFS [length + 3].

7. Next, each element of the FFS array starting from -length


+ 1 element and up to 0 is calculated by the formula:

FFS [-i] = a * DFS [-i + 2] + b * DFS [-i + 3] + d * FFS [-i +


1] + c * DSF [-i + 4] + e * FFS [ -i + 2].

8. We remove 2 service elements from the FFS array and get


a filtered signal.
The block diagram of the algorithm is shown in Figure 2.
The block diagram of the Butterworth filter algorithm implementation

1119

Authorized licensed use limited to: University of Durham. Downloaded on November 01,2020 at 14:32:42 UTC from IEEE Xplore. Restrictions apply.
2020 International Russian Automation Conference (RusAutoCon)
IV. EXPERIMENTAL RESEARCH During the experiment, measurements were made at each of
To conduct experimental studies, measurements were taken their positions, shown in Figure 4. As a result of each
of the RSSI signal level in a pre-marked room. The floor plan is measurement, a multidimensional time series was obtained,
shown in Figure 3. The following equipment was used for this: consisting of 4 data series - one for each BLE beacon.

1. A mobile device for reading RSSI signal strength After obtaining the time series (Figure 1), the following data
indications based on the Android operating system. processing was performed:

2. BLE beacons based on the specification of Bluetooth 5.0. 1. The average signal value was found - which was taken as
the reference value.
3. A personal computer for collecting, processing and
visualizing the information received. 2. The entire time series was shifted along the OY axis by
the average value (Figure 5).

Time series of noise

3. The remaining data can be mistaken for noise and it is


necessary to establish the nature of this noise. To do this, a
histogram of the noise distribution is constructed (Figure 6),
which reflects the law of its distribution.

Floor plan

BLE beacons were placed along the perimeter of the room


and 4 measurement zones were marked. The layout of
lighthouses and zones is shown in Figure 4.

Noise distribution law

4. As a result, the Butterworth filter is applied to the time


series (Figure 7).

Butterworth filter operation example

V. CONCLUSIONS
In conclusion, we can conclude that the Butterworth filter
can be used to filter signals from BLE beacons, but the main
problem is that the filter requires a large sample, and an increase
in the number of values in the time series of the signal will slow
down the filtering process.

The layout of lighthouses and zones

1120

Authorized licensed use limited to: University of Durham. Downloaded on November 01,2020 at 14:32:42 UTC from IEEE Xplore. Restrictions apply.
2020 International Russian Automation Conference (RusAutoCon)
This work was financially supported by the Ministry of [9] Le Thanh andPo Orten, “Enhancing Accuracy Performance of Bluetooth
Science and Higher Education of the Russian Federation (State Positioning,” Enhancing Accuracy Performance of Bluetooth Positioning,
pp. 2726–2731, 2007. DOI:10.1109/WCNC.2007.506.
task of VlSU GB-1187/20).
[10] Julang Ying, Kaveh Pahlavan, and Xinrong Li, “Precision of RSS-based
indoor geolocation in IoT applications,” Personal Indoor and Mobile
REFERENCES Radio Communications, pp. 1–5, 2017.
[1] Development of the digital economy in Russi, program until 2035. [11] A.Vashistha, A. Gupta, and C. L. Law, “Self calibration of the anchor
[Online]. Available: http://innclub.info/wp- nodes for UWB-IR TDOA based indoor positioning system,” IEEE 4th
content/uploads/2017/05/strategy.pdf World Forum on Internet of Things, 2018. DOI: 10.1109/wf-
[2] Internet of Things in 2020 roadmap to future. [Online]. Available: iot.2018.8355163
https://mafiadoc.com/internet-of-things-in-2020-roadmap-for-the-future- [12] V.M.Grinyak,T.M. Grinyak, andP.A. Tsybanov,“Positioning indoors
eposs_5a17c5b41723ddb59566366a.html using Bluetooth devices,” Territory of new features, no. 2(41), 2018.
[3] N. L. Scott, B. Hansen, C. A. LaDue, and C. Lam, “Using an active Radio [Online]. Available: https://cyberleninka.ru/article/n/pozitsionirovanie-
Frequency Identification Real-Time Location System to remotely monitor vnutri-pomescheniy-s-pomoschyu-bluetooth-ustroystv
animal movement in zoos,” Animal Biotelemetry, 2016. DOI: [13] M.A.Youssef,A.Agrawala, andA. Udaya Shankar, “WLAN location
10.1186/s40317-016-0108-5. determination via clustering and probability distributions,” WLAN
[4] I. Angulo, E. Onieva, A. Perallos, and I. Salaberria, “Low Cost Real Time location determination via clustering and probability distributions, pp.
Location System Based in Radio Frequency Identification for the 143–150, 2003. DOI: 10.1109/PERCOM.2003.1192736.
Provision of Social and Safety Services,” Wireless Personal [14] Navigine – navigation and movement tracking inside buildings.[Online].
Communications, vol. 84, is. 4, pp. 2797–2814, 2015. DOI: Available: https://vc.ru/tribuna/11212-navigine
10.1007/s11277-015-2767-6.
[15] R.Xu,W. Chen,Y. Xu, and S. Ji, “A New Indoor Positioning System
[5] P.Agarwal,A. Gupta,G. Verma, andH. Verma, “Wireless Monitoring and Architecture Using GPS Signals,” Sensors, 15, pp. 10074–10087, 2015.
Indoor Navigation of a Mobile Robot Using RFID,” inNature Inspired
Computing. Advances in Intelligent Systems and Computing, vol 652,B. [16] Amir Tabatabaei and MohammadReza,“Performance analysis of
Panigrahi,M. Hoda, andV. Sharma, Eds.Singapore:Springer, 2018. DOI: GLONASS integration with GPS vectorised receiver in urban canyon
10.1007/978-981-10-6747-1_10 positioning,” Survey Review. DOI: 10.1080/00396265.2018.1481181
[6] Eric J. M. Gunther, Levin J. Sliker,and Cathy Bodine, “A UHF RFID [17] M.S.Tigina andE.D. Pavlintseva,“The use of technology Beacon for
positioning system for use in warehouse navigation by employees with information navigation within the premises,” Bulletin of MGUP,no. 1,
cognitive disability,” Disability and Rehabilitation: Assistive 2015. [Online]. Available: https://cyberleninka.ru/article/n/primenenie-
Technology, 12:8, pp. 832–842, 2017. DOI: tehnologii-beacon-dlya-osuschestvleniya-informatsionnoy-navigatsii-v-
10.1080/17483107.2016.1274342 predelah-pomescheniya
[7] Y. Zhuang, Z. Syed, Y. Li, and N. El-Sheimy, “Evaluation of Two WiFi [18] Peter C. Karlsson, “Sensor Fused Indoor Positioning Using Dual Band
Positioning Systems Based on Autonomous Crowdsourcing of Handheld WiFi Signal Measurements,” MSc Thesis, Department of Automatic
Devices for Indoor Navigation,” IEEE Transactions on Mobile Control, Lund University, 2014.
Computing, vol. 15, no. 8, pp. 1982–1995, 2016. [19] Johan Svenér, Peter C. Karlsson, and Magnus Persson, “Indoor location
[8] Juan Luo, Zhenyan Zhang, Chang Liu, and Haibo Luo, “Reliable and detection using combined radio frequency signal and audio frequency
Cooperative Target Tracking Based on WSN and WiFi in Indoor Wireless signal beacon system,” USA Patent US9955315B2, 2015.
Networks,” Access IEEE, vol. 6, pp. 24846–24855, 2018.

1121

Authorized licensed use limited to: University of Durham. Downloaded on November 01,2020 at 14:32:42 UTC from IEEE Xplore. Restrictions apply.

You might also like