- Email: [email protected]
Mobile user tracking system with ZigBee
Accepted Manuscript
Mobile user tracking system with ZigBee Jacek Stepien´ , Jacek Kołodziej , Witold Machowski PII: DOI: Reference:
S0141-9331(16)00038-7 10.1016/j.micpro.2016.02.007 MICPRO 2353
To appear in:
Microprocessors and Microsystems
Received date: Revised date: Accepted date:
20 October 2015 7 January 2016 5 February 2016
Please cite this article as: Jacek Stepien´ , Jacek Kołodziej , Witold Machowski , Mobile user tracking system with ZigBee, Microprocessors and Microsystems (2016), doi: 10.1016/j.micpro.2016.02.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Jacek STĘPIEŃ, Jacek KOŁODZIEJ, Witold MACHOWSKI AGH University of Science and Technology Krakow, Department of Electronics 30 Adama Mickiewicza Av, 30-059 Krakow, Poland e-mail: [jacek.stepien, jacek.kolodziej, witold.machowski]@agh.edu.pl
Mobile user tracking system with ZigBee Abstract. In the paper an implementation of mobile nodes tracking system based on ZigBee and Wi-Fi wireless networks is presented. On the base of known algorithmic as well as circuit solutions a simple yet universal system, applied in prototype application dedicated for person’s localization in museum premises has been developed. Since system utilizes entirely wireless communication, it can be applied in any closed objects. The system has been preliminarily verified in real in-situ environment.
CR IP T
Keywords: ZigBee, tracking systems, localization algorithm, embedded systems, microprocessors. Introduction
Stationary node 1.2
Coordinator11 Coordinator
Stationary node 2.1
Stationary node 2.1
ED
Mobile Node
M
AN US
Museums as well as art galleries are often very specific venues. They may be historical buildings, where no installation work is allowed (museums) or space re-arrangements such as significant lighting or sound system modification (art galleries). On the other hand modern objects of that kind require – for different reasons – to be equipped with smart monitoring and control allowing for instance an automatic detection of visitors in specific room and/or supervising the group size. When systems with portable audio guides are used, two basic requirements should be fulfilled: Wireless communication between stationary and mobile system’s elements, which should assure that the specific information reaches the group of visitors (respectively its guide) independently on their position. For obvious reason this communication system must cover entire building, Possibility of tracking for chosen mobile elements (e.g. groups’ guides). On one hand it may guarantee an optimal visitors groups traffic to avoid eventual jams, supplying the group with appropriate information related with its position (e.g. information abort the object, the group just is nearby). On the other, it may be used to activate additional building automation elements such as extra lighting or muting the music in appropriate region – i.e where the group stays at.
ZigBee Network II
Stationary node 1.1
PT
ZigBee Network I
Room 1
Coordinator 2
Room 2
AC
CE
Wi-Fi network
Stationary node 3.1
Main system
ZigBee Network III
Stationary node 3.2
Coordinator 3
Room 3
Fig. 1. System structure
Widespread use of different wireless communication protocols simplifies an adjustment of new solutions to specific requirements of an application under the design. In the presented system one of fundamental assumptions was to assure entirely wireless communication between all elements: stationary, firmly located in predefined points as well as mobile ones all persons moving within the venue are equipped with. Simultaneously stationary nodes must exhibit some flexibility – allowing simple spatial reconfiguration of the system, when necessary. Moreover due to battery operation of mobile nodes
ACCEPTED MANUSCRIPT
CR IP T
energy consumption must be optimized, otherwise frequent recharge is necessary. Finally, functional requirements are different for different node types. In the system four types of nodes have to cooperate with each other: Main (coordinating) nodes – located in each room. It collects the information form all the other nodes in particular room and communicates with the central system; Stationary nodes – each room is equipped with at least two such nodes (bigger or unusual rooms – e.g. that with untypical shape may need even more nodes of that type). They are used for tracking mobile nodes, sending to mobile nodes data related to their actual positions and driving room amenities. Mobile nodes – facilities born by group guides. They communicate with stationary- and main nodes in order to properly locate the group within the venue. Secondary mobile nodes – all visitors belonging to specific group have them. That nodes can only receive data and play voice messages. Actually that kind of nodes are not a part of the system – their operation is local and limited to the area where the information of visible object is broadcasted. Due to system specification and significant functionality differentiation for particular nodes we decided to use different processors and controllers fitting different requirements for each node type, two independent wireless communication protocols were also used. First protocol, used by main nodes to communicate with central computer is the IEEE802.11 (Wi-Fi) protocol, the second used by stationary and mobile nodes is ZigBee. The structure of the system is shown in Fig.1.
AN US
Assumptions about system operation are as follows: The group enters the room – coordinator node detects a mobile node, an information about room occupation is sent to the central system, which blocks the possibility of coming another group (just single group may be in one room at the moment); Stationary nodes perform cyclic measurements of object position Gathered information are sent to the central system it eventually activates additional elements (light, background music, relevant information for secondary mobile nodes) accordingly to object position; Group leave the room – information is sent to the central system, which releases the blocking process, next group may enter the room. In the paper we present first development stage related with implementation of mobile user localization and sending its position coordinates to the central system. The second implemented functionality is broadcasting the general information addressed to entire group. ZigBee protocol
AC
CE
PT
ED
M
Over the last dozen years both researchers and practitioners are more and more interested in short distance wireless transmission. Actually we all have been observing significant paradigm change – what previously was wireless (e.g. long distance range radio transmission) is being substituted by “wire” (fiber optics – like backbone networks) and vice-versa (keyboards, mousse, headphones, printers etc.). For computer networks that type of transmission at the very beginning was considered as an alternative to wired transmission, treated as better and more robust. Therefore wireless transmission was considered as an ultimate choice. Both devices as well as protocols were immature, consequently most wireless solutions had significantly worse transmission parameters than cable ones. However later on, protocols have been refined, advances in electron devices technology resulted with significant reliability improvement, and nowadays wireless dominates in short range data transmission. More and more application areas are discovered. In local areas networks (LANs) for personal computers new variants of IEEE 802.11 protocol appeared, some personal area network (PAN) have been elaborated with the most popular Bluetooth (IEEE 802.15.1) standard, the same for long distance wireless (GSM, IEEE 802.16). However that protocols share one specific feature – they are relatively sophisticated, with expanded data validation and protection procedures, excluding or at least making very difficult energy optimization for the transmission process. Though wireless sensor networks require as long as possible mobile sensor nodes operation (otherwise frequent recharging their batteries is necessary). For needs of sensor networks, as well as home automation networks many wireless communication protocols have been developed. They are suitable for low-power, short range (100m max) and relatively low data throughput (up to 100kbps). One of such a standard is ZigBee. It belongs to LR-WPAN (Low-Rate Wireless Personal Area Network) class. ZigBee standard defines recommendations for network as well as application layer, utilizing for direct data transmission specified by IEEE 802.15.4 standard. Both recommendations are to be treated as single entity, defining rules for transmission. IEEE 802.15.4 standard was developed to fit the needs low-bit rate and short range transmission. Devices conforming the standard are low-power. Standard describes physical layer as well as medium access control (MAC) sublayer of ISO-OSI model. Basic features of that protocol comprise:
wireless data transmission with: 20, 40, 100 or 250 kbps rates; supports different network topologies: star, peer-to-peer or mesh; two addressing modes; with short 16-bit or extended 64-bit; CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) protocol; low energy consumption; possibility of signal strength monitoring (RSSI – Received Signal Strength Indicator); channel quality detection - LQI (Link Quality Indication); choice of frequency channel used for transition [1].
ACCEPTED MANUSCRIPT Two or more devices communicating with each other via transmission over shared physical channel create WPAN. This network must have a supervisor device being the coordinator. The other devices, depending on their functionality may work as end nodes (only communicating with parent device) or routers, which re-transmit data to subnets. Master nodes because of their dedication to serve very different tasks, usually operate connected to AC power grid, while end nodes are predisposed to operate with battery supply. Routers are suited to work as normal slave nodes and their additional functionality is to retransmit data. Therefore supply scheme for the system are to be adjusted to specific needs of particular application.
CR IP T
ZigBee standard defines implementation rules for higher network layers (i.e. application) and recommendations about cooperation with MAC sub-layer. These are: network creation rules; attaching and detaching devices; configuration rules for newly attached nodes; identification of neighbor nodes; managing the mechanism guaranteed time slots (GTS); format of transmission frame; routing of messages; trace recognition [2].
ED
M
AN US
The ZigBee protocol chosen for implementation of course has alternatives among other low energy communication protocols. In recent years Low Energy Bluetooth (BLE) is also used for sensor networks, on the other hand the ANT protocol became more and more popular. Ranking of energy efficiency of all aforementioned protocols however is still disputable and there is no clear and unquestionable statement of that issue among the researchers. The transceivers’ current consumption for all protocols is comparable one with another and fits into 16-20 mA for transmit/receive mode and about 2-5μA in sleep mode. The complexity of communication protocol structure is also similar, the same apply for energy dissipation related with maintenance (initialization, registration, network ID broadcasting), but with that aspect both ZigBee as well as ANT seem to have some advantage over sophisticated BLE protocol. However from energy efficiency perspective all protocols seem to be similar. The argument pro BLE or ANT+ might be that some modern smartphones support them and have appropriate interfaces. Nevertheless for application under the question that feature may be a significant drawback – visitors may bear such devices, what can intrude the operation of localization mechanisms, moreover leading to significant increase the power consumption by the nodes. The ultimate reasons justifying the ZiGBee choice were: significantly wider real transmission range compared to BLE and ANT (generally speaking ZigBee is an network protocol, while the rest two belong to WPAN class); possibility of implementing mesh topology and support of multi-stage transmission (not supported by BLE) – which is provided in the next development phase; lower sensitivity for possible interference with WiFi network - with ZigBee a transmission in separated frequency channels may be forced (BLE, which utilizes a frequency hopping does not support that feature popularity and low price of transmission modules (main advantage over ANT) [3]
PT
It should be noted that communication protocol layer implement mechanisms of strength and quality of received signal. Based on that it is possible to evaluate quality of data transmission, or as in our application – to determine the position of mobile object. The RSSI index expressed in dBm determines the power level of received signal, while LQI describes with parameters the quality of received data (most frequently using RSSI and information about error rate level in the network).
AC
CE
An obvious issue related to wireless transmission are interferences coming from the fact of sharing ISM 2.4 GHz band. From the considered application perspective was to guarantee a coexistence of sensor nodes network (utilizing ZigBee transmissions) and backbone network nodes transmitting data with IEEE802.11 standard. The minimization of interference was achieved by appropriate configuration of used devices working in mutually separated frequency channels. As a result the only potential intruders are mobile phones and their peripherals with active Bluetooth interfaces. This kind of interference are for sure possible in the system, however they are not very harmful. Due to frequency hopping which is principal for Bluetooth the duration of interference noise from BT intruding ZigBee as well as WiFi transmission channel is relatively short, so transmissions not necessarily overlap. Moreover for developed system time intervals utilized for localization purposes are of at least 1s duration, so even if transmission is disturbed and single packet is lost it could be successfully retransmitted. Principles of mobile node positioning There are elaborated many mobile node localization methods for wireless networks. Most accurate of them require significant expansion of node structure. It could be either precise time synchronization for nodes, when time of transmission (ToT) is measured [4] or directional antennas implementing angle of arrival (AoA) method [5,6]. The simplicity of system under construction forces use less precise, yet assuring acceptable accuracy methods of localization. They are based on measurement of the power of received signal. In ZigBee network most frequently RSSI is used (but sometimes LQI) [6-9]. That method class could be further divided for direct and indirect methods. In direct methods a power of received signal is measured and node coordinates are determined with previously pre-defined power map – prepared in test phase. For indirect methods, measured power level is converted into another value first, and only then localization algorithm is realized.
ACCEPTED MANUSCRIPT One of the most used indirect methods is to determine approximate object location based on the power received from test node, and then refine the result with the aid of mathematical methods [10-20]. In the literature on the subject [10,17] a Friis equation is usually being referenced as most elementary formula relating distance between nodes and power damping level. Formula assumes ideal conditions for transmission and propagation for the system comprising just single transmitting and single receiving antenna. In practice there is just single case in which Friis formula can be utilized without significant error – i.e. satellite communication.The fundamental quantitative description of received signal power given by Friis formula is: 2
(1)
where: PR-power of received signal, PT-power of transmitted signal, GR-gain of receiving antenna, GT-gain of transmitting antenna, l – wavelength, d – distance between transmitter and receiver.
CR IP T
PR l GR GT PT 4d
However above formula has no practical importance, because ideal condition of wave transmission and propagation are assumed. In practice, for determining the distance between nodes the formula is applied:
PR P0 10n log
d d0
(2)
AN US
with: PR-power of received signal, P0-power of signal received by receiver at i d0 distance from the transmitter, n –damping coefficient.
Parametr n is a parameter that describe how the signal strength decreases when the distance from the transmitter increases. The value of n is highly dependent of the environment (the wall thickness will influent a lot). So that this value can only be determined empirically
M
Equation (2) despite significant simplification allows for relatively high measurement accuracy. The main requirement is performing initial calibration (to estimate P0 value) individually for each receiver and environmental condition similar to actual one. In systems with enhanced accuracy measurement of incoming signal power is done from at least three reference nodes. On that basis rough approximation of distance of mobile node to stationary ones is done, next refinement using triangulation methods is performed. Basic triangulation method is centroid localization (CI). In measurement setup stationary reference nodes with known coordinates Ni N j ( x, y) have to be defined. Position of mobile node within
1 n
n
N (x , y ) j
j
j
j 1
(3)
PT
M i ( xi , yi )
ED
coverage of stationary node transmitters is determined based on relationship:
CE
Imperfection of that method is clearly visible at a glance – the result of calculation is position of mobile node in equal distance to all the reference nodes. Moreover, possibilities offered by the system are not utilized with full extend, because only the fact of mobile node presence within coverage of stationary node is taken into account, while the power of received signal is ignored. Therefore enhanced algorithm has been elaborated implementing so called weighted centroid localization (WCL) – contribution of each stationary node in location the object is corrected by additional weight factor, which is related to received signal power. Now position of mobile nodes is calculated from: n
w N (x , y )
AC
ij
M i ( xi , yi )
j
j
j 1
n
w
j
(4)
ij
j 1
Where:
wij
1 (d ij ) k
determines the share of particular node in location of mobile node;
dij – denotes the distance between mobile and stationary node (determined on the basis of measured power of received signal – RSSI); while k describes weight factor attributed to particular node during location procedure. Reference [9] describes experimental results obtained using that method. Localization in the system
ACCEPTED MANUSCRIPT Basic assumption of designed localization network is the use of three immovable nodes with known mutual distances. The structure of network cell in single room is depicted in Fig.2. Master node (being in the same time network coordinator) gathers information about the signal strength (RSSI) from stationary and mobile nodes.
N3=(x3,y3) r3 M=(x0,y0)
r1
r2 N2=(x2,y2)
CR IP T
N1=(x1,y1)
Fig. 2.Localization of mobile node principle
AN US
To determine coordinates of mobile node the equation of the circle with radius r and center at P(x0,y0) is considered:
r 2 ( x x0 )2 ( y y0 )2
Coordinates of subsequent nodes in the system are defined as: M = (x0,y0) – mobile node, N1 = (x1,y1) – reference node #1 (master), N2 = (x2,y2) – reference node #2, N3 = (x3,y3) – reference node #3, r1, r2, r3 – radii of circles defining location of the mobile node.
M
With no loss of generality we can assume, that master node is in the origin of the reference frame i.e. (x1,y1) = (0,0) Thus solution of the system of equation:
x 2 y 2 r12 ( x x2 )2 ( y y2 ) 2 r22 2
(5)
ED
( x x3 ) ( y y3 ) 2
r32
Structure of the system
PT
gives coordinates of the mobile node.
AC
CE
Structure (fragmentary) of the system is shown in Fig. 1. Each from dozen or so rooms is equipped with the coordinator and at least two stationary nodes. On the other hand if their number exceeds 3 (it may result from the need for dedicated building automation control systems), during a pre-configuration phase three nodes are selected to be responsible for localization purposes. That choice is determined by position of the nodes as well as propagation conditions characteristic for each room. One of stationary nodes (usually that situated nearby the room entrance) is the master. Master node manages the network operation – attaches remaining stationary nodes and mobile node), realizes cyclic RSSI measurements for each stationary nodes in room, receives information about RSSI coefficients of mobile node and organizes entire communication with central system. Communication between mobile nodes, stationary ones and the master is realized using ZigBee protocol. Additionally all coordinating nodes create main transmission network of the system. All are equipped with communication modules Wi-Fi (IEEE 802.11g) used for data are sent to main system switchboard. For multiple functionality and relatively extended structure coordinator nodes are always supplied with AC electricity line.
ACCEPTED MANUSCRIPT RSSI[dBm] -80
-70
-40
-20
Distance [cm]
CR IP T
Fig. 2. Experimental data for RSSI coefficient versus distance for two nodes
xi
x p1 x p 2 x p3 3
, yi
y p1 y p 2 y p3
(6)
3
AN US
Remaining stationary nodes, configured as ZigBee network router, realize cyclic transmissions to coordinator. This allows (on the basis of RSSI and LQI) evaluate the transmission quality in the room; they also deliver information about RSSI level for the signal received from mobile node. Additionally, dependently on purpose they can realize extra functions – driving auxiliary equipment as lighting and/or loudspeakers. They may have battery or AC line supply. Mobile node is an miniature device with battery supply. Just after attaching to the new sub-network (room) perform cyclic signal transmissions – which are used to determine RSSI. It also allows voice communication with a group guide (sending a short voice message from the centre). Mobile node is configured as a ZigBee router. In the framework of experiments, later used during system creation and configuration tests of efficiency of node localization within single room have been performed [21,22]. Tests purpose was verification of possible use of selected methods for credible localization results. Obtained results allow to approximate RSSI versus distance relationship with the aid of linear function (Fig. 3). Measurement error of RSSI has however a big impact. In extremes it may cause (5) to be contradictory. To avoid the case when proper calculations of object being localized is impossible an approximate value of coordinates of mobile node is calculated. A relationships for calculating centre of gravity for homogenous triangle was used for that purpose:
AC
CE
PT
ED
M
with M = (xi,yi) – mobile node coordinates (see Fig. 4 for details). One of fundamental problems when implementing that method was variation of RSSI coefficient with orientation of transmitter module. The simplicity of transmitter antenna (copper strips at the PCB) results with significant change of measured factor with any turn of the device. Obtained measurement results [21] made possible directional characterization describing RSSI change as a function of turn angle of transmitter antenna. Data are collected in table 1. Measured coefficient variability is so high, that RSSI to distance conversion error may extend in extreme cases 50%. In preliminary phase that problem was solved by supervised distribution of communication modules adjusting the signal emission angle. The next problem appearing when using wireless transmission systems are possible Wi-Fi/ZigBee interferences, since both systems share common ISM 2.4 GHz band. To minimize mutual interference possibility coordinating modules supervise the transmission process in order to avoid simultaneous transmission of both systems by sharing predefined timeslots.
N3=(x3,y3) p3=(xp3,yp3) p1=(xp1,yp1) M=(x0,y0) N1=(x1,y1) p2=(xp2,yp2) N2=(x2,y2)
Fig. 4. Finding approximate location of mobile node
ACCEPTED MANUSCRIPT Angle[deg] / Distance [m] 0.5 1 2 2.5
0
Table 1. RSSI dependence on turn angle of ATZB-24 module 45 90 135 180 225
-79 -82 -88 -91-fault
-81 -78 -83 -90
-73 -78 -89 -88
-73 -77 -82 -86
-74 -81 -84 -85
80 -78 -84 -89
270
315
-79 -89 -88 -87
-73 -88 -87 -91-fault
Battery power supply
AN US
CR IP T
Communication in ZigBee network was realized with ATMEL ATZB-24-A2 (ZigBit) [23] modules. They comprise simple transmitting/receiving antenna and single-chip Atmega1281 microcontroller with ZigBee protocol stack (BitCloud). An user can also write own program into internal memory. This program realize not only ZigBee transmission but also other tasks using microcontroller peripherals. It simplifies the design and system architecture, because e.g. mobile node practically comprise ZigBit module with supplying elements only. It is very important in energy minimization context. Similarly, stationary nodes have simple architecture. In basic form they have exactly the same functionality as mobile nodes, the only difference is supplying scheme – either battery or AC line power supply. However, as mentioned before they should be suited to perform additional functions as activating auto-guide, lighting etc. If stationary node functionality is expanded, the microcontroller from ZigBit module is not able to serve entire functionality any more. In that case the node is implemented with independent additional microcontroller (ATMEL ATmega32) with ZigBee module connected with UART interface. The device comprise also driving elements respectively to functionality provided for the device. Stationary nodes with extended functionality are always supplied from AC electricity. The most sophisticated node is coordinator. Besides communicating with other systems in the room using ZigBee protocol it realizes additional steering as other stationary nodes. Additionally it gathers the information about signal level from all remaining nodes and sends it to central system. Coordinator circuits and central unit communicate with each other using Wi-Fi (IEEE802.11). Because coordinator must be suited for multi-task and multi-thread operation that module was implemented with ARM-core STM32F4 microcontroller. The device realizes massive data transfer and assures appropriate transmission parameters for communication with central unit. Wi-Fi communication was based on RS-9110 [24] modules from Redpine Signals company. Block diagrams of used devices are shown in Fig. 5.
ATZB-24R (ZigBit)
Battery power supply
ATZB-24R (ZigBit)
ED
``
M
a)
AC power supply
AtMega32 Microcontroller
b)
AC power supply
Control components
ATZB-24R (ZigBit)
RS9110 WLAN
STM32F4 Microcontroller
Control components
c)
PT
Fig. 5. Block diagrams of equipment used in our system a) mobile node, b) stationary node, c) supervisor
AC
CE
Principle of system operation is as described below: In initial state in the room a local ZigBee network supervised by a coordinator is created. Supervisor receives from stationary nodes information about current RSSI coefficients. Based on that, a correctness of the information about mutual distances between stationary nodes and coordinator (that distances are saved in the central unit) is verified. This verification simultaneously serves for evaluating the quality of transmission in ZigBee. When mobile node enters the room ZigBee network coordinator attaches mobile node to the net. Next the information about new node being localized is sent to all the stationary nodes, which are configured as ZigBee routers. Coordinator and stationary nodes communicate periodically with mobile node and measure RSSI coefficient. Measured RSSIs are transmitted to coordinator module, which sends them to the central unit. Central unit determines mobile node position and via coordinator module sends data to drive periphery devices. When system detects that mobile node leaved the room, module is logged-out from the ZigBee network and central unit is notified that next mobile node (group) may enter, if awaits such permission. In the idle state, when no mobile node is in, central unit via coordinating module may send/receive data to and from stationary modules. Fig. 6 presents the algorithm of system coordinator operation.
ACCEPTED MANUSCRIPT Start
Cooordinaor Initizlization procedures
Register Mobile Node
Network Initialization
Send information to Stationary Nodes
Stationary Nodes registration
Collect RSSI information from Stationary Nodes
Stationary Nodes RSSI Measurement
Send Data to Central Komputer
CR IP T
Delay Send Data to Central Computer
Receive Data from Central Computer and send to Stationary and Mobile Nodes
Delay Send Network Invitation
Yes
No
Yes
Is answer From Mobile Node?
Is Mobile Node still in touch?
No
AN US
Remove Mobile Node from Network Send Information tro Central System
Fig. 6. Algorithm of system coordinator operation
CE
PT
ED
M
Power consumption estimation of designed system is not a simple matter because of asynchronous event-based transmission. A total power consumption by nodes depends on the number of mobile nodes being localized within a time interval, localization rate and number of (auto) retransmissions. We estimate current consumption during the most power saving phases of node operation (i.e. during transmitting receiving data). The maximum current in mobile node does not exceed 20mA, while for stationary nodes (without extended functionality) 22 mA. In Figs. 7-9 photographs of most important system components are shown.
AC
Fig. 7 Prototype of the mobile node
ACCEPTED MANUSCRIPT
AN US
CR IP T
Fig. 8 Prototype of the coordinator
M
Fig 9. Assembled stationary node with omnidirectional antenna
ED
Future work
Summary
CE
PT
Developed simple prototype of the system realizes localization processes and communication with central control unit. Distortions observed during tests resulting from incorrect readout of RSSI cause that future research will focus on assuring better reliability and accuracy of the system. To improve that transmission modules will be equipped with omnidirectional antennas. Further development of supervising software will also be necessary, to protect against simultaneous ZigBee and Wi-Fi transmission. Moreover system will be expanded with an extra stationary nodes (at least three are necessary in each room) to improve localization procedures.
AC
In the paper a system of automated identification of group guides in museum has been presented. The system identifies the mobile object using localization algorithms in a network working with ZigBee network. A feasibility study has been performed. Preliminary research results verified capabilities as well as limitations of implemented localization algorithms together with chosen communication modules. Next the physical prototype has been build. This practical implementation confirmed possible operation of such a system, limitations related to inaccuracy of realized distance measurements were also shown. Preliminary in-situ measurements proved that measurement accuracy increases with number of reference nodes. Elaborated system is not sufficiently mature so far, to be used as an independent back-end application. Nevertheless the experience gained within the preliminary phase and obtained results are promising and sufficient to plan further improved evaluation version of the system in question. Acknowledgement The work presented in the paper has been supported by AGH University of Science and Technology, Department of Electronics under statutory activity. REFERENCES
ACCEPTED MANUSCRIPT [1] IEEE Computer Society. Part 15.4: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (WPANs). IEEE Std 802.15.4TM-2006. [2] ZigBee Standards Organization. ZigBee Specification. Document 053474r17 [3] Stepień J., Kołodziej J., Machowski W. - Low Energy Personal Wireless Sensor Networks (in Polish) Proc. of The 12th Conference on Selected Problems of Electrical Engineering and Electronics WZEE’2015, [4] Schwarzer S., Vossiek M., Pichler M., Stelzer A, Precise Distance Measurement with IEEE 802.15.4 (ZigBee) DevicesRadio and Wireless Symposium, 2008 [5] Hyuntae Cho, Hyunsung Jang, Yunju Baek. Practical Localization System for Consumer Devices using Zigbee Networks, IEEE Transactions on Consumer Electronics (Volume:56 , Issue: 3) [6] Chul Young Park, DaeHeon Park, JangWoo Park, YangSun Lee, Youngeun An, Localization algorithm design and implementation to utilization RSSI and AoA of Zigbee, IEEE, 2010 [7] Kamol Kaemarungsi, Rachasak Ranron, Prasit Pongsoon, Study of Received Signal Strength Indication in ZigBee Location Cluster for Indoor Localization, IEEE, 2013
CR IP T
[8] Swagatika Bohidar, Sasmita Behera, Chitta Ranjan Tripathy, A Comparative View on Received Signal Strength (RSS) Based location Estimation in WSN, 2015 IEEE International Conference on Engineering and Technology (ICETECH), 20th March 2015, Coimbatore, TN, India [9] Jia Chen, Xiao-jun Wu, Feng Ye, Ping Song, and Jian-wei Liu, Improved RSSI based Localization Algorithm for Park Lighting Control and Child Location Tracking, Information and Automation, 2009. ICIA '09 [10] Blumenthal, J.; Grossmann, R.; Golatowski, F.; Timmermann, D. Weighted Centroid Localization in Zigbee-based Sensor Networks, IEEE International Symposium on Intelligent Signal Processing WISP 2007, 1-6 [11] Domınguez-Duran M., Claros D., Urdiales C., Coslado F., Sandoval F., Dynamic calibration and zero configuration positioning system for WSN, Electrotechnical Conference, MELECON 2008
AN US
[12] Yong-Zheng Li, Lei Wang, Xiao-Ming Wu, Yuan-Ting Zhang, Experimental Analysis on Radio Transmission and Localization of a Zigbee-based Wireless Healthcare Monitoring Platform, Information Technology and Applications in Biomedicine, 2008 [13] Yibin Lin, Juan Li, Yanmin Zhu, Cross-platform Interference, Locating WiFi Devices with ZigBee Sensor Networks, The first International Workshop on Quality of Experience-Based Wireless Communications, IEEE, 2015 [14] Navarro-Alvarez E., Siller M., A Node Localization Scheme for Zigbee-based Sensor Networks, Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics, San Antonio, TX, USA - October 2009 [15] Ng Yew Keong, Kuek Ser Chieh, Muhamad Fadli Burhan, Nagaletchumi Balasubramaniam, Norashidah Md. Din, RFID and ZigBee Integrated Environment for Indoor Localization, 2014 4th International Conference on Engineering Technology and Technopreneuship
M
[16] Mendalka M., Kulas Ł., Nyka K., Localization in wireless sensor networks based on ZigBee platform, Microwaves, Radar and Wireless Communications, 2008, MIKON 2008
ED
[17] Mendalka, M.; Bizewski, K.; Kulas, L.; Nyka, K., Pattern Matching Localization in ZigBee Wireless Sensor Networks, 18th International Conference on Microwave Radar and Wireless Communications (MIKON), 2010, 1-4 [17] Xin Hu, Lianglun Cheng,Guangchi Zhang, A Zigbee-based localization Algorithm For Indoor Environments, 2011 International Conference on Computer Science and Network Technology
PT
[18] Shengnan Gai, Eui-Jung Jung, Byung-Ju Yi, Localization Algorithm Based on Zigbee Wireless Sensor Network with Application to an Active Shopping Cart, 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2014) September 14-18, 2014, Chicago, IL, USA [19] Angela Song-Ie Noh, Woong Jae Lee, Jin Young Ye, Comparison of the mechanisms of the Zigbee’s indoor localization algorithm, Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, 2008
CE
[20] Wenping Chen, Xiaofeng Meng, A Cooperative Localization Scheme for ZigBee-based Wireless Sensor Networks, Networks, 2006. ICON '06 [21] Oleksy M., Tracking system with ZigBee modules (in Polish), BSc Project supervised by Dr Jacek Stępień, AGH University and Technology, 2013
AC
[22] Nikiel P., Localization methods of moving nodes in ZigBee networks (in Polish), BSc Project supervised by Dr Jacek Kołodziej, AGH University and Technology, 2013 [23] Atmel Corporation. ZigBitTM 2.4 GHz Wireless Modules - ATZB-24-A2/B0 –Datasheet [24] Redpine Signals Incorporation, RS-9110 Self Contained 802.11 b/g/n Module with Networking Stack – Datasheet
Biography
ACCEPTED MANUSCRIPT
AN US
CR IP T
Jacek Stepien ([email protected]) received M.Sc and PhD degrees both in electronics from AGH University of Science and Technology in Krakow, Poland in 1992 and 2001 respectively. Since 1992 he is with AGH Department of Electronics. His professional activity focuses nowadays on communication system, especially wireless sensor networks protocols with energy optimization. He is author and co-author of over 70 scientific papers.
AC
CE
PT
ED
M
Jacek Kolodziej ([email protected]) received the M.Sc in Electronics and Telecommunication in 1999 and PhD (with honors) degree in Electronics in 2007, both from the AGH University of Science and Technology Krakow, Poland. His scientific interests include advanced electronic circuits in telecommunication, wireless sensor networks, analog – digital converters, data synchronization, time and frequency transfer, software engineering, testing and reliability. He is co-author of 45 conference and journal papers. He is IEEE member, IEEE Young Processional Chair, Student Branch Chancellor. He has been participating in several R&D Polish and European projects being personally responsible for system management and communication procedures.
Witold Machowski ([email protected]) received M.Sc (electronics), PhD (physics) DSc (electronics) degrees from AGH University of Science and Technology in Kraków, Poland in 1986, 1993 and 2013 respectively. Since 1992 he has been with AGH Department of Electronics. His professional activity focuses on microelectronics – mainly analog and mixed-signal CMOS. He is author and co-author of over 90 scientific papers and 3 books. Between 1999 and 2005 he held a position of Deputy Dean for Teaching at AGH UST Faculty of Electrical Engineering, Automatics, Computer Science and Electronics. He is also a IEEE member and PTETiS (Polish Society for Theoretical and Applied Electrical Engineering) executive board member.
Mobile user tracking system with ZigBee Jacek Stepien´ , Jacek Kołodziej , Witold Machowski PII: DOI: Reference:
S0141-9331(16)00038-7 10.1016/j.micpro.2016.02.007 MICPRO 2353
To appear in:
Microprocessors and Microsystems
Received date: Revised date: Accepted date:
20 October 2015 7 January 2016 5 February 2016
Please cite this article as: Jacek Stepien´ , Jacek Kołodziej , Witold Machowski , Mobile user tracking system with ZigBee, Microprocessors and Microsystems (2016), doi: 10.1016/j.micpro.2016.02.007
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT Jacek STĘPIEŃ, Jacek KOŁODZIEJ, Witold MACHOWSKI AGH University of Science and Technology Krakow, Department of Electronics 30 Adama Mickiewicza Av, 30-059 Krakow, Poland e-mail: [jacek.stepien, jacek.kolodziej, witold.machowski]@agh.edu.pl
Mobile user tracking system with ZigBee Abstract. In the paper an implementation of mobile nodes tracking system based on ZigBee and Wi-Fi wireless networks is presented. On the base of known algorithmic as well as circuit solutions a simple yet universal system, applied in prototype application dedicated for person’s localization in museum premises has been developed. Since system utilizes entirely wireless communication, it can be applied in any closed objects. The system has been preliminarily verified in real in-situ environment.
CR IP T
Keywords: ZigBee, tracking systems, localization algorithm, embedded systems, microprocessors. Introduction
Stationary node 1.2
Coordinator11 Coordinator
Stationary node 2.1
Stationary node 2.1
ED
Mobile Node
M
AN US
Museums as well as art galleries are often very specific venues. They may be historical buildings, where no installation work is allowed (museums) or space re-arrangements such as significant lighting or sound system modification (art galleries). On the other hand modern objects of that kind require – for different reasons – to be equipped with smart monitoring and control allowing for instance an automatic detection of visitors in specific room and/or supervising the group size. When systems with portable audio guides are used, two basic requirements should be fulfilled: Wireless communication between stationary and mobile system’s elements, which should assure that the specific information reaches the group of visitors (respectively its guide) independently on their position. For obvious reason this communication system must cover entire building, Possibility of tracking for chosen mobile elements (e.g. groups’ guides). On one hand it may guarantee an optimal visitors groups traffic to avoid eventual jams, supplying the group with appropriate information related with its position (e.g. information abort the object, the group just is nearby). On the other, it may be used to activate additional building automation elements such as extra lighting or muting the music in appropriate region – i.e where the group stays at.
ZigBee Network II
Stationary node 1.1
PT
ZigBee Network I
Room 1
Coordinator 2
Room 2
AC
CE
Wi-Fi network
Stationary node 3.1
Main system
ZigBee Network III
Stationary node 3.2
Coordinator 3
Room 3
Fig. 1. System structure
Widespread use of different wireless communication protocols simplifies an adjustment of new solutions to specific requirements of an application under the design. In the presented system one of fundamental assumptions was to assure entirely wireless communication between all elements: stationary, firmly located in predefined points as well as mobile ones all persons moving within the venue are equipped with. Simultaneously stationary nodes must exhibit some flexibility – allowing simple spatial reconfiguration of the system, when necessary. Moreover due to battery operation of mobile nodes
ACCEPTED MANUSCRIPT
CR IP T
energy consumption must be optimized, otherwise frequent recharge is necessary. Finally, functional requirements are different for different node types. In the system four types of nodes have to cooperate with each other: Main (coordinating) nodes – located in each room. It collects the information form all the other nodes in particular room and communicates with the central system; Stationary nodes – each room is equipped with at least two such nodes (bigger or unusual rooms – e.g. that with untypical shape may need even more nodes of that type). They are used for tracking mobile nodes, sending to mobile nodes data related to their actual positions and driving room amenities. Mobile nodes – facilities born by group guides. They communicate with stationary- and main nodes in order to properly locate the group within the venue. Secondary mobile nodes – all visitors belonging to specific group have them. That nodes can only receive data and play voice messages. Actually that kind of nodes are not a part of the system – their operation is local and limited to the area where the information of visible object is broadcasted. Due to system specification and significant functionality differentiation for particular nodes we decided to use different processors and controllers fitting different requirements for each node type, two independent wireless communication protocols were also used. First protocol, used by main nodes to communicate with central computer is the IEEE802.11 (Wi-Fi) protocol, the second used by stationary and mobile nodes is ZigBee. The structure of the system is shown in Fig.1.
AN US
Assumptions about system operation are as follows: The group enters the room – coordinator node detects a mobile node, an information about room occupation is sent to the central system, which blocks the possibility of coming another group (just single group may be in one room at the moment); Stationary nodes perform cyclic measurements of object position Gathered information are sent to the central system it eventually activates additional elements (light, background music, relevant information for secondary mobile nodes) accordingly to object position; Group leave the room – information is sent to the central system, which releases the blocking process, next group may enter the room. In the paper we present first development stage related with implementation of mobile user localization and sending its position coordinates to the central system. The second implemented functionality is broadcasting the general information addressed to entire group. ZigBee protocol
AC
CE
PT
ED
M
Over the last dozen years both researchers and practitioners are more and more interested in short distance wireless transmission. Actually we all have been observing significant paradigm change – what previously was wireless (e.g. long distance range radio transmission) is being substituted by “wire” (fiber optics – like backbone networks) and vice-versa (keyboards, mousse, headphones, printers etc.). For computer networks that type of transmission at the very beginning was considered as an alternative to wired transmission, treated as better and more robust. Therefore wireless transmission was considered as an ultimate choice. Both devices as well as protocols were immature, consequently most wireless solutions had significantly worse transmission parameters than cable ones. However later on, protocols have been refined, advances in electron devices technology resulted with significant reliability improvement, and nowadays wireless dominates in short range data transmission. More and more application areas are discovered. In local areas networks (LANs) for personal computers new variants of IEEE 802.11 protocol appeared, some personal area network (PAN) have been elaborated with the most popular Bluetooth (IEEE 802.15.1) standard, the same for long distance wireless (GSM, IEEE 802.16). However that protocols share one specific feature – they are relatively sophisticated, with expanded data validation and protection procedures, excluding or at least making very difficult energy optimization for the transmission process. Though wireless sensor networks require as long as possible mobile sensor nodes operation (otherwise frequent recharging their batteries is necessary). For needs of sensor networks, as well as home automation networks many wireless communication protocols have been developed. They are suitable for low-power, short range (100m max) and relatively low data throughput (up to 100kbps). One of such a standard is ZigBee. It belongs to LR-WPAN (Low-Rate Wireless Personal Area Network) class. ZigBee standard defines recommendations for network as well as application layer, utilizing for direct data transmission specified by IEEE 802.15.4 standard. Both recommendations are to be treated as single entity, defining rules for transmission. IEEE 802.15.4 standard was developed to fit the needs low-bit rate and short range transmission. Devices conforming the standard are low-power. Standard describes physical layer as well as medium access control (MAC) sublayer of ISO-OSI model. Basic features of that protocol comprise:
wireless data transmission with: 20, 40, 100 or 250 kbps rates; supports different network topologies: star, peer-to-peer or mesh; two addressing modes; with short 16-bit or extended 64-bit; CSMA/CA (Carrier Sense Multiple Access with Collision Avoidance) protocol; low energy consumption; possibility of signal strength monitoring (RSSI – Received Signal Strength Indicator); channel quality detection - LQI (Link Quality Indication); choice of frequency channel used for transition [1].
ACCEPTED MANUSCRIPT Two or more devices communicating with each other via transmission over shared physical channel create WPAN. This network must have a supervisor device being the coordinator. The other devices, depending on their functionality may work as end nodes (only communicating with parent device) or routers, which re-transmit data to subnets. Master nodes because of their dedication to serve very different tasks, usually operate connected to AC power grid, while end nodes are predisposed to operate with battery supply. Routers are suited to work as normal slave nodes and their additional functionality is to retransmit data. Therefore supply scheme for the system are to be adjusted to specific needs of particular application.
CR IP T
ZigBee standard defines implementation rules for higher network layers (i.e. application) and recommendations about cooperation with MAC sub-layer. These are: network creation rules; attaching and detaching devices; configuration rules for newly attached nodes; identification of neighbor nodes; managing the mechanism guaranteed time slots (GTS); format of transmission frame; routing of messages; trace recognition [2].
ED
M
AN US
The ZigBee protocol chosen for implementation of course has alternatives among other low energy communication protocols. In recent years Low Energy Bluetooth (BLE) is also used for sensor networks, on the other hand the ANT protocol became more and more popular. Ranking of energy efficiency of all aforementioned protocols however is still disputable and there is no clear and unquestionable statement of that issue among the researchers. The transceivers’ current consumption for all protocols is comparable one with another and fits into 16-20 mA for transmit/receive mode and about 2-5μA in sleep mode. The complexity of communication protocol structure is also similar, the same apply for energy dissipation related with maintenance (initialization, registration, network ID broadcasting), but with that aspect both ZigBee as well as ANT seem to have some advantage over sophisticated BLE protocol. However from energy efficiency perspective all protocols seem to be similar. The argument pro BLE or ANT+ might be that some modern smartphones support them and have appropriate interfaces. Nevertheless for application under the question that feature may be a significant drawback – visitors may bear such devices, what can intrude the operation of localization mechanisms, moreover leading to significant increase the power consumption by the nodes. The ultimate reasons justifying the ZiGBee choice were: significantly wider real transmission range compared to BLE and ANT (generally speaking ZigBee is an network protocol, while the rest two belong to WPAN class); possibility of implementing mesh topology and support of multi-stage transmission (not supported by BLE) – which is provided in the next development phase; lower sensitivity for possible interference with WiFi network - with ZigBee a transmission in separated frequency channels may be forced (BLE, which utilizes a frequency hopping does not support that feature popularity and low price of transmission modules (main advantage over ANT) [3]
PT
It should be noted that communication protocol layer implement mechanisms of strength and quality of received signal. Based on that it is possible to evaluate quality of data transmission, or as in our application – to determine the position of mobile object. The RSSI index expressed in dBm determines the power level of received signal, while LQI describes with parameters the quality of received data (most frequently using RSSI and information about error rate level in the network).
AC
CE
An obvious issue related to wireless transmission are interferences coming from the fact of sharing ISM 2.4 GHz band. From the considered application perspective was to guarantee a coexistence of sensor nodes network (utilizing ZigBee transmissions) and backbone network nodes transmitting data with IEEE802.11 standard. The minimization of interference was achieved by appropriate configuration of used devices working in mutually separated frequency channels. As a result the only potential intruders are mobile phones and their peripherals with active Bluetooth interfaces. This kind of interference are for sure possible in the system, however they are not very harmful. Due to frequency hopping which is principal for Bluetooth the duration of interference noise from BT intruding ZigBee as well as WiFi transmission channel is relatively short, so transmissions not necessarily overlap. Moreover for developed system time intervals utilized for localization purposes are of at least 1s duration, so even if transmission is disturbed and single packet is lost it could be successfully retransmitted. Principles of mobile node positioning There are elaborated many mobile node localization methods for wireless networks. Most accurate of them require significant expansion of node structure. It could be either precise time synchronization for nodes, when time of transmission (ToT) is measured [4] or directional antennas implementing angle of arrival (AoA) method [5,6]. The simplicity of system under construction forces use less precise, yet assuring acceptable accuracy methods of localization. They are based on measurement of the power of received signal. In ZigBee network most frequently RSSI is used (but sometimes LQI) [6-9]. That method class could be further divided for direct and indirect methods. In direct methods a power of received signal is measured and node coordinates are determined with previously pre-defined power map – prepared in test phase. For indirect methods, measured power level is converted into another value first, and only then localization algorithm is realized.
ACCEPTED MANUSCRIPT One of the most used indirect methods is to determine approximate object location based on the power received from test node, and then refine the result with the aid of mathematical methods [10-20]. In the literature on the subject [10,17] a Friis equation is usually being referenced as most elementary formula relating distance between nodes and power damping level. Formula assumes ideal conditions for transmission and propagation for the system comprising just single transmitting and single receiving antenna. In practice there is just single case in which Friis formula can be utilized without significant error – i.e. satellite communication.The fundamental quantitative description of received signal power given by Friis formula is: 2
(1)
where: PR-power of received signal, PT-power of transmitted signal, GR-gain of receiving antenna, GT-gain of transmitting antenna, l – wavelength, d – distance between transmitter and receiver.
CR IP T
PR l GR GT PT 4d
However above formula has no practical importance, because ideal condition of wave transmission and propagation are assumed. In practice, for determining the distance between nodes the formula is applied:
PR P0 10n log
d d0
(2)
AN US
with: PR-power of received signal, P0-power of signal received by receiver at i d0 distance from the transmitter, n –damping coefficient.
Parametr n is a parameter that describe how the signal strength decreases when the distance from the transmitter increases. The value of n is highly dependent of the environment (the wall thickness will influent a lot). So that this value can only be determined empirically
M
Equation (2) despite significant simplification allows for relatively high measurement accuracy. The main requirement is performing initial calibration (to estimate P0 value) individually for each receiver and environmental condition similar to actual one. In systems with enhanced accuracy measurement of incoming signal power is done from at least three reference nodes. On that basis rough approximation of distance of mobile node to stationary ones is done, next refinement using triangulation methods is performed. Basic triangulation method is centroid localization (CI). In measurement setup stationary reference nodes with known coordinates Ni N j ( x, y) have to be defined. Position of mobile node within
1 n
n
N (x , y ) j
j
j
j 1
(3)
PT
M i ( xi , yi )
ED
coverage of stationary node transmitters is determined based on relationship:
CE
Imperfection of that method is clearly visible at a glance – the result of calculation is position of mobile node in equal distance to all the reference nodes. Moreover, possibilities offered by the system are not utilized with full extend, because only the fact of mobile node presence within coverage of stationary node is taken into account, while the power of received signal is ignored. Therefore enhanced algorithm has been elaborated implementing so called weighted centroid localization (WCL) – contribution of each stationary node in location the object is corrected by additional weight factor, which is related to received signal power. Now position of mobile nodes is calculated from: n
w N (x , y )
AC
ij
M i ( xi , yi )
j
j
j 1
n
w
j
(4)
ij
j 1
Where:
wij
1 (d ij ) k
determines the share of particular node in location of mobile node;
dij – denotes the distance between mobile and stationary node (determined on the basis of measured power of received signal – RSSI); while k describes weight factor attributed to particular node during location procedure. Reference [9] describes experimental results obtained using that method. Localization in the system
ACCEPTED MANUSCRIPT Basic assumption of designed localization network is the use of three immovable nodes with known mutual distances. The structure of network cell in single room is depicted in Fig.2. Master node (being in the same time network coordinator) gathers information about the signal strength (RSSI) from stationary and mobile nodes.
N3=(x3,y3) r3 M=(x0,y0)
r1
r2 N2=(x2,y2)
CR IP T
N1=(x1,y1)
Fig. 2.Localization of mobile node principle
AN US
To determine coordinates of mobile node the equation of the circle with radius r and center at P(x0,y0) is considered:
r 2 ( x x0 )2 ( y y0 )2
Coordinates of subsequent nodes in the system are defined as: M = (x0,y0) – mobile node, N1 = (x1,y1) – reference node #1 (master), N2 = (x2,y2) – reference node #2, N3 = (x3,y3) – reference node #3, r1, r2, r3 – radii of circles defining location of the mobile node.
M
With no loss of generality we can assume, that master node is in the origin of the reference frame i.e. (x1,y1) = (0,0) Thus solution of the system of equation:
x 2 y 2 r12 ( x x2 )2 ( y y2 ) 2 r22 2
(5)
ED
( x x3 ) ( y y3 ) 2
r32
Structure of the system
PT
gives coordinates of the mobile node.
AC
CE
Structure (fragmentary) of the system is shown in Fig. 1. Each from dozen or so rooms is equipped with the coordinator and at least two stationary nodes. On the other hand if their number exceeds 3 (it may result from the need for dedicated building automation control systems), during a pre-configuration phase three nodes are selected to be responsible for localization purposes. That choice is determined by position of the nodes as well as propagation conditions characteristic for each room. One of stationary nodes (usually that situated nearby the room entrance) is the master. Master node manages the network operation – attaches remaining stationary nodes and mobile node), realizes cyclic RSSI measurements for each stationary nodes in room, receives information about RSSI coefficients of mobile node and organizes entire communication with central system. Communication between mobile nodes, stationary ones and the master is realized using ZigBee protocol. Additionally all coordinating nodes create main transmission network of the system. All are equipped with communication modules Wi-Fi (IEEE 802.11g) used for data are sent to main system switchboard. For multiple functionality and relatively extended structure coordinator nodes are always supplied with AC electricity line.
ACCEPTED MANUSCRIPT RSSI[dBm] -80
-70
-40
-20
Distance [cm]
CR IP T
Fig. 2. Experimental data for RSSI coefficient versus distance for two nodes
xi
x p1 x p 2 x p3 3
, yi
y p1 y p 2 y p3
(6)
3
AN US
Remaining stationary nodes, configured as ZigBee network router, realize cyclic transmissions to coordinator. This allows (on the basis of RSSI and LQI) evaluate the transmission quality in the room; they also deliver information about RSSI level for the signal received from mobile node. Additionally, dependently on purpose they can realize extra functions – driving auxiliary equipment as lighting and/or loudspeakers. They may have battery or AC line supply. Mobile node is an miniature device with battery supply. Just after attaching to the new sub-network (room) perform cyclic signal transmissions – which are used to determine RSSI. It also allows voice communication with a group guide (sending a short voice message from the centre). Mobile node is configured as a ZigBee router. In the framework of experiments, later used during system creation and configuration tests of efficiency of node localization within single room have been performed [21,22]. Tests purpose was verification of possible use of selected methods for credible localization results. Obtained results allow to approximate RSSI versus distance relationship with the aid of linear function (Fig. 3). Measurement error of RSSI has however a big impact. In extremes it may cause (5) to be contradictory. To avoid the case when proper calculations of object being localized is impossible an approximate value of coordinates of mobile node is calculated. A relationships for calculating centre of gravity for homogenous triangle was used for that purpose:
AC
CE
PT
ED
M
with M = (xi,yi) – mobile node coordinates (see Fig. 4 for details). One of fundamental problems when implementing that method was variation of RSSI coefficient with orientation of transmitter module. The simplicity of transmitter antenna (copper strips at the PCB) results with significant change of measured factor with any turn of the device. Obtained measurement results [21] made possible directional characterization describing RSSI change as a function of turn angle of transmitter antenna. Data are collected in table 1. Measured coefficient variability is so high, that RSSI to distance conversion error may extend in extreme cases 50%. In preliminary phase that problem was solved by supervised distribution of communication modules adjusting the signal emission angle. The next problem appearing when using wireless transmission systems are possible Wi-Fi/ZigBee interferences, since both systems share common ISM 2.4 GHz band. To minimize mutual interference possibility coordinating modules supervise the transmission process in order to avoid simultaneous transmission of both systems by sharing predefined timeslots.
N3=(x3,y3) p3=(xp3,yp3) p1=(xp1,yp1) M=(x0,y0) N1=(x1,y1) p2=(xp2,yp2) N2=(x2,y2)
Fig. 4. Finding approximate location of mobile node
ACCEPTED MANUSCRIPT Angle[deg] / Distance [m] 0.5 1 2 2.5
0
Table 1. RSSI dependence on turn angle of ATZB-24 module 45 90 135 180 225
-79 -82 -88 -91-fault
-81 -78 -83 -90
-73 -78 -89 -88
-73 -77 -82 -86
-74 -81 -84 -85
80 -78 -84 -89
270
315
-79 -89 -88 -87
-73 -88 -87 -91-fault
Battery power supply
AN US
CR IP T
Communication in ZigBee network was realized with ATMEL ATZB-24-A2 (ZigBit) [23] modules. They comprise simple transmitting/receiving antenna and single-chip Atmega1281 microcontroller with ZigBee protocol stack (BitCloud). An user can also write own program into internal memory. This program realize not only ZigBee transmission but also other tasks using microcontroller peripherals. It simplifies the design and system architecture, because e.g. mobile node practically comprise ZigBit module with supplying elements only. It is very important in energy minimization context. Similarly, stationary nodes have simple architecture. In basic form they have exactly the same functionality as mobile nodes, the only difference is supplying scheme – either battery or AC line power supply. However, as mentioned before they should be suited to perform additional functions as activating auto-guide, lighting etc. If stationary node functionality is expanded, the microcontroller from ZigBit module is not able to serve entire functionality any more. In that case the node is implemented with independent additional microcontroller (ATMEL ATmega32) with ZigBee module connected with UART interface. The device comprise also driving elements respectively to functionality provided for the device. Stationary nodes with extended functionality are always supplied from AC electricity. The most sophisticated node is coordinator. Besides communicating with other systems in the room using ZigBee protocol it realizes additional steering as other stationary nodes. Additionally it gathers the information about signal level from all remaining nodes and sends it to central system. Coordinator circuits and central unit communicate with each other using Wi-Fi (IEEE802.11). Because coordinator must be suited for multi-task and multi-thread operation that module was implemented with ARM-core STM32F4 microcontroller. The device realizes massive data transfer and assures appropriate transmission parameters for communication with central unit. Wi-Fi communication was based on RS-9110 [24] modules from Redpine Signals company. Block diagrams of used devices are shown in Fig. 5.
ATZB-24R (ZigBit)
Battery power supply
ATZB-24R (ZigBit)
ED
``
M
a)
AC power supply
AtMega32 Microcontroller
b)
AC power supply
Control components
ATZB-24R (ZigBit)
RS9110 WLAN
STM32F4 Microcontroller
Control components
c)
PT
Fig. 5. Block diagrams of equipment used in our system a) mobile node, b) stationary node, c) supervisor
AC
CE
Principle of system operation is as described below: In initial state in the room a local ZigBee network supervised by a coordinator is created. Supervisor receives from stationary nodes information about current RSSI coefficients. Based on that, a correctness of the information about mutual distances between stationary nodes and coordinator (that distances are saved in the central unit) is verified. This verification simultaneously serves for evaluating the quality of transmission in ZigBee. When mobile node enters the room ZigBee network coordinator attaches mobile node to the net. Next the information about new node being localized is sent to all the stationary nodes, which are configured as ZigBee routers. Coordinator and stationary nodes communicate periodically with mobile node and measure RSSI coefficient. Measured RSSIs are transmitted to coordinator module, which sends them to the central unit. Central unit determines mobile node position and via coordinator module sends data to drive periphery devices. When system detects that mobile node leaved the room, module is logged-out from the ZigBee network and central unit is notified that next mobile node (group) may enter, if awaits such permission. In the idle state, when no mobile node is in, central unit via coordinating module may send/receive data to and from stationary modules. Fig. 6 presents the algorithm of system coordinator operation.
ACCEPTED MANUSCRIPT Start
Cooordinaor Initizlization procedures
Register Mobile Node
Network Initialization
Send information to Stationary Nodes
Stationary Nodes registration
Collect RSSI information from Stationary Nodes
Stationary Nodes RSSI Measurement
Send Data to Central Komputer
CR IP T
Delay Send Data to Central Computer
Receive Data from Central Computer and send to Stationary and Mobile Nodes
Delay Send Network Invitation
Yes
No
Yes
Is answer From Mobile Node?
Is Mobile Node still in touch?
No
AN US
Remove Mobile Node from Network Send Information tro Central System
Fig. 6. Algorithm of system coordinator operation
CE
PT
ED
M
Power consumption estimation of designed system is not a simple matter because of asynchronous event-based transmission. A total power consumption by nodes depends on the number of mobile nodes being localized within a time interval, localization rate and number of (auto) retransmissions. We estimate current consumption during the most power saving phases of node operation (i.e. during transmitting receiving data). The maximum current in mobile node does not exceed 20mA, while for stationary nodes (without extended functionality) 22 mA. In Figs. 7-9 photographs of most important system components are shown.
AC
Fig. 7 Prototype of the mobile node
ACCEPTED MANUSCRIPT
AN US
CR IP T
Fig. 8 Prototype of the coordinator
M
Fig 9. Assembled stationary node with omnidirectional antenna
ED
Future work
Summary
CE
PT
Developed simple prototype of the system realizes localization processes and communication with central control unit. Distortions observed during tests resulting from incorrect readout of RSSI cause that future research will focus on assuring better reliability and accuracy of the system. To improve that transmission modules will be equipped with omnidirectional antennas. Further development of supervising software will also be necessary, to protect against simultaneous ZigBee and Wi-Fi transmission. Moreover system will be expanded with an extra stationary nodes (at least three are necessary in each room) to improve localization procedures.
AC
In the paper a system of automated identification of group guides in museum has been presented. The system identifies the mobile object using localization algorithms in a network working with ZigBee network. A feasibility study has been performed. Preliminary research results verified capabilities as well as limitations of implemented localization algorithms together with chosen communication modules. Next the physical prototype has been build. This practical implementation confirmed possible operation of such a system, limitations related to inaccuracy of realized distance measurements were also shown. Preliminary in-situ measurements proved that measurement accuracy increases with number of reference nodes. Elaborated system is not sufficiently mature so far, to be used as an independent back-end application. Nevertheless the experience gained within the preliminary phase and obtained results are promising and sufficient to plan further improved evaluation version of the system in question. Acknowledgement The work presented in the paper has been supported by AGH University of Science and Technology, Department of Electronics under statutory activity. REFERENCES
ACCEPTED MANUSCRIPT [1] IEEE Computer Society. Part 15.4: Wireless Medium Access Control (MAC) and Physical Layer (PHY) Specifications for Low-Rate Wireless Personal Area Networks (WPANs). IEEE Std 802.15.4TM-2006. [2] ZigBee Standards Organization. ZigBee Specification. Document 053474r17 [3] Stepień J., Kołodziej J., Machowski W. - Low Energy Personal Wireless Sensor Networks (in Polish) Proc. of The 12th Conference on Selected Problems of Electrical Engineering and Electronics WZEE’2015, [4] Schwarzer S., Vossiek M., Pichler M., Stelzer A, Precise Distance Measurement with IEEE 802.15.4 (ZigBee) DevicesRadio and Wireless Symposium, 2008 [5] Hyuntae Cho, Hyunsung Jang, Yunju Baek. Practical Localization System for Consumer Devices using Zigbee Networks, IEEE Transactions on Consumer Electronics (Volume:56 , Issue: 3) [6] Chul Young Park, DaeHeon Park, JangWoo Park, YangSun Lee, Youngeun An, Localization algorithm design and implementation to utilization RSSI and AoA of Zigbee, IEEE, 2010 [7] Kamol Kaemarungsi, Rachasak Ranron, Prasit Pongsoon, Study of Received Signal Strength Indication in ZigBee Location Cluster for Indoor Localization, IEEE, 2013
CR IP T
[8] Swagatika Bohidar, Sasmita Behera, Chitta Ranjan Tripathy, A Comparative View on Received Signal Strength (RSS) Based location Estimation in WSN, 2015 IEEE International Conference on Engineering and Technology (ICETECH), 20th March 2015, Coimbatore, TN, India [9] Jia Chen, Xiao-jun Wu, Feng Ye, Ping Song, and Jian-wei Liu, Improved RSSI based Localization Algorithm for Park Lighting Control and Child Location Tracking, Information and Automation, 2009. ICIA '09 [10] Blumenthal, J.; Grossmann, R.; Golatowski, F.; Timmermann, D. Weighted Centroid Localization in Zigbee-based Sensor Networks, IEEE International Symposium on Intelligent Signal Processing WISP 2007, 1-6 [11] Domınguez-Duran M., Claros D., Urdiales C., Coslado F., Sandoval F., Dynamic calibration and zero configuration positioning system for WSN, Electrotechnical Conference, MELECON 2008
AN US
[12] Yong-Zheng Li, Lei Wang, Xiao-Ming Wu, Yuan-Ting Zhang, Experimental Analysis on Radio Transmission and Localization of a Zigbee-based Wireless Healthcare Monitoring Platform, Information Technology and Applications in Biomedicine, 2008 [13] Yibin Lin, Juan Li, Yanmin Zhu, Cross-platform Interference, Locating WiFi Devices with ZigBee Sensor Networks, The first International Workshop on Quality of Experience-Based Wireless Communications, IEEE, 2015 [14] Navarro-Alvarez E., Siller M., A Node Localization Scheme for Zigbee-based Sensor Networks, Proceedings of the 2009 IEEE International Conference on Systems, Man, and Cybernetics, San Antonio, TX, USA - October 2009 [15] Ng Yew Keong, Kuek Ser Chieh, Muhamad Fadli Burhan, Nagaletchumi Balasubramaniam, Norashidah Md. Din, RFID and ZigBee Integrated Environment for Indoor Localization, 2014 4th International Conference on Engineering Technology and Technopreneuship
M
[16] Mendalka M., Kulas Ł., Nyka K., Localization in wireless sensor networks based on ZigBee platform, Microwaves, Radar and Wireless Communications, 2008, MIKON 2008
ED
[17] Mendalka, M.; Bizewski, K.; Kulas, L.; Nyka, K., Pattern Matching Localization in ZigBee Wireless Sensor Networks, 18th International Conference on Microwave Radar and Wireless Communications (MIKON), 2010, 1-4 [17] Xin Hu, Lianglun Cheng,Guangchi Zhang, A Zigbee-based localization Algorithm For Indoor Environments, 2011 International Conference on Computer Science and Network Technology
PT
[18] Shengnan Gai, Eui-Jung Jung, Byung-Ju Yi, Localization Algorithm Based on Zigbee Wireless Sensor Network with Application to an Active Shopping Cart, 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS 2014) September 14-18, 2014, Chicago, IL, USA [19] Angela Song-Ie Noh, Woong Jae Lee, Jin Young Ye, Comparison of the mechanisms of the Zigbee’s indoor localization algorithm, Software Engineering, Artificial Intelligence, Networking, and Parallel/Distributed Computing, 2008
CE
[20] Wenping Chen, Xiaofeng Meng, A Cooperative Localization Scheme for ZigBee-based Wireless Sensor Networks, Networks, 2006. ICON '06 [21] Oleksy M., Tracking system with ZigBee modules (in Polish), BSc Project supervised by Dr Jacek Stępień, AGH University and Technology, 2013
AC
[22] Nikiel P., Localization methods of moving nodes in ZigBee networks (in Polish), BSc Project supervised by Dr Jacek Kołodziej, AGH University and Technology, 2013 [23] Atmel Corporation. ZigBitTM 2.4 GHz Wireless Modules - ATZB-24-A2/B0 –Datasheet [24] Redpine Signals Incorporation, RS-9110 Self Contained 802.11 b/g/n Module with Networking Stack – Datasheet
Biography
ACCEPTED MANUSCRIPT
AN US
CR IP T
Jacek Stepien ([email protected]) received M.Sc and PhD degrees both in electronics from AGH University of Science and Technology in Krakow, Poland in 1992 and 2001 respectively. Since 1992 he is with AGH Department of Electronics. His professional activity focuses nowadays on communication system, especially wireless sensor networks protocols with energy optimization. He is author and co-author of over 70 scientific papers.
AC
CE
PT
ED
M
Jacek Kolodziej ([email protected]) received the M.Sc in Electronics and Telecommunication in 1999 and PhD (with honors) degree in Electronics in 2007, both from the AGH University of Science and Technology Krakow, Poland. His scientific interests include advanced electronic circuits in telecommunication, wireless sensor networks, analog – digital converters, data synchronization, time and frequency transfer, software engineering, testing and reliability. He is co-author of 45 conference and journal papers. He is IEEE member, IEEE Young Processional Chair, Student Branch Chancellor. He has been participating in several R&D Polish and European projects being personally responsible for system management and communication procedures.
Witold Machowski ([email protected]) received M.Sc (electronics), PhD (physics) DSc (electronics) degrees from AGH University of Science and Technology in Kraków, Poland in 1986, 1993 and 2013 respectively. Since 1992 he has been with AGH Department of Electronics. His professional activity focuses on microelectronics – mainly analog and mixed-signal CMOS. He is author and co-author of over 90 scientific papers and 3 books. Between 1999 and 2005 he held a position of Deputy Dean for Teaching at AGH UST Faculty of Electrical Engineering, Automatics, Computer Science and Electronics. He is also a IEEE member and PTETiS (Polish Society for Theoretical and Applied Electrical Engineering) executive board member.