Extending TETRA with wireless sensor networks

Int. J. Intelligent Engineering Informatics, Vol. X, No. Y, xxxx

1

Extending TETRA with wireless sensor networks Mario Paoli*, Francesco Ficarola, Ugo Maria Colesanti and Andrea Vitaletti

Comment [A1]: Author: Please confirm if M. Paoli is the corresponding author.

Department of Computer, Automatic and Management Engineering, Sapienza – University of Rome, Italy Email: [email protected] Email: [email protected] Email: [email protected] Email: [email protected] *Corresponding author

Simona Citrigno and Domenico Saccà Centro di Competenza Tecnologico CCT ICT-SUD, Università della Calabria, Italy Email: [email protected] Email: [email protected] Abstract: The Terrestrial Trunked Radio (TETRA) system is an open standard developed by ETSI and designed to support mobile radio communications in a number of market segments, among which public safety is by far the largest one. In this paper, we present the activities to the ‘TETRis – TETRA Innovative Open Source Services’ project to integrate wireless sensor networks technology with TETRA, in order to support real-time feedback in two application contexts relevant in public safety scenarios: structural health monitoring and air quality monitoring. The WSN deployed in our testbed is based on the MagoNode platform, a new mote featuring an RF front-end capable to enhance the radio performance. The results of the experimental activity confirm that WSNs can be effectively used to support the management of critical situations in the considered scenarios and that the MagoNode platform well meet the requirements provided by experts on structural health monitoring. Keywords: wireless sensor networks; WSN; Terrestrial Trunked Radio; TETRA; TETRis; structural health monitoring; environmental monitoring; embedded systems. Reference to this paper should be made as follows: Paoli, M., Ficarola, F., Colesanti, U.M., Vitaletti, A., Citrigno, S. and Saccà, D. (xxxx) ‘Extending TETRA with wireless sensor networks’, Int. J. Intelligent Engineering Informatics, Vol. X, No. Y, pp.xxx–xxx. Biographical notes:

Copyright © 20XX Inderscience Enterprises Ltd.

Comment [A2]: Author: Please provide the biographical details of no more than 100 words for each author.

2

M. Paoli et al.

This paper is a revised and expanded version of a paper entitled ‘MagoNode: advantages of RF front-ends in wireless sensor networks’ presented at Workshop on Real-World Wireless Sensor (RealWSN), Como, Italy, 19–20 September 2013.

1

Introduction

The Terrestrial Trunked Radio (TETRA, http://www.etsi.org/technologiesclusters/ technologies/tetra) is an open standard for mobile radio communications developed by the European Telecommunications Standards Institute (ETSI, http://www.etsi.org/) and specifically designed to support professional mobile radio communications in a number of market segments such as public safety, transportation, utilities, government, military,

Comment [A3]: Author: Please check and confirm the conference statement is correct.

Extending TETRA with wireless sensor networks

3

commerce, industry. TETRA is deployed in over 88 countries worldwide, and the main market is by far the public safety. While the initial focus of TETRA has been on voice communications, data communications have been supported since the beginning and nowadays are gaining more importance. Indeed, the prevention and management of critical situations related to public safety requires the real-time acquisition of data from the field, to react more consciously and faster to dangerous events and to greatly increase the safety of the practitioners. Wireless sensor networks (WSNs) (see Akyildiz and Vuran, 2010; Stankovic, 2008) are a state-of-the-art technology to collect real-time data on the field. They are ad hoc networks made of battery powered devices able to communicate wirelessly and to collect data on a phenomenon of interest, as shown by Hill et al. (2004). Since nodes are battery powered and they communicate over the wireless medium, WSNs can be easily and cost effectively deployed in public safety scenarios, as pointed out by Mainwaring et al. (2002). The video by the RUNESEU project (available at http://youtu.be/ RU21YO6XF_o) well-motivates the employment of WSNs to improve the safety of critical infrastructures. In that video, a fire takes place in a tunnel, and the incident area network (i.e., a WSN) provides relevant information to the rescue operators that are supported in their activities by a network that could reasonably be TETRA-like. In this paper, we present the efforts made in the context of the ‘TETRis – TETRA Innovative Open Source Services’ (http://www.ponrec.it/opendata/risultati/ricercaindustriale/pon01_00451/) project, with the purpose of enriching the public safety services and applications running on top of the TETRA communication system, with real-time data collected in the field by a new WSN platform. The remainder of this paper is organised as follows: in Section 2, we give more details on TETRA and the TETRis project. Section 3 describes the MagoNode, the new low-power WSN platform employed in our testbed and compares its performance with state-of-the-art competitors. In Section 3.3, we present the hardware interfaces that have been implemented for the MagoNode to collect data of interest in the two reference scenarios considered in this paper, namely structural health monitoring and air quality monitoring. Finally, in Section 4, we discuss the experimental results on the performance of a WSN made of MagoNode motes in a real testbed also employed as one of the final demonstrators of the TETRis project, in order to show that our platform is suitable for fast, easy, reliable and long-lasting deployment on public safety scenarios.

2

TETRA network and the TETRIS project

The TETRA (http://www.etsi.org/technologiesclusters/technologies/tetra) is a digital trunked mobile radio standard developed for the needs of professional mobile radio-communication. It is defined by the ETSI (http://www.etsi.org/). TETRA market includes among the possible clients, public security forces, agencies operating in emergencies contexts, military forces, hospitals, but also organisations responsible for the public services. The TETRA standard follows the evolution of the digital communication for public networks (i.e., GSM, UMTS), even if with a slower timetable, that assures the reliability of consolidated standards required by the markets to whom TETRA is directed.

4

M. Paoli et al.

The main point of strength of the TETRA communication standard is the much lower frequency used in comparison with other mobile communications standards such as GSM. That allows to achieve wider geographic area coverage with a smaller number of transmitters, thus significantly improving the cost effectiveness of the overall infrastructure. Furthermore, unlike traditional cellular networks in which only one-to-one communications are supported, TETRA can support one-to-one, one-to-many and many-to-many communications with high security standards and in the absence of a network, terminals can be used as walkie-talkies. All these feature, makes TETRA particularly well-suited for the management of critical situation and public safety scenarios that are described in details in the TETRA Critical Communication Association (TCCA) website (http://www.tandcca.com/). Here, we only mention the successful employment of TETRA to guarantee the safety of remarkable events such as the FIFAWorld CUP 2010 in South Africa and the Bejing Olympics Games in China. In these events, and more in general in events in which a number of public safety forces – police, fire fighters, ambulance, etc. – have to be coordinated, TETRA provides a common and standard infrastructure for secure and reliable communications. Among TETRA competitors, we considered Digital Mobile Radio (DMR, http://www.etsi.org/technologies-clusters/technologies/digitalmobile-radio). Like TETRA, DMR is developed by ETSI and it is designed for professional, commercial and private radio users. While TETRA and DMR provide similar performance, benefits offered by TETRA in terms of spectrum efficiency, data throughput and power efficiency outperform those offered by DMR. However, DMR has wider coverage than TETRA that is beneficial in rural areas deployments (see TETRA vs. DMR, 2012). While the initial focus of TETRA has been on voice communications, data communications have been supported since the beginning and nowadays are gaining more importance. Data transfer is efficient and long range (many km), but slow by modern standards with at most 691.2 kbit/s in an expanded 150 kHz channel. While this constraint limits the applicability of TETRA for advanced multimedia and interactive functionalities, it fully supports most of the application scenarios foreseen for WSNs, in which the traffic generated by the sensors is usually in the order of few kbit/s (most common radio transceivers support at most 250 kbit/s). The real-time information collected by the WSNs are used to react more consciously and faster to dangerous events and to greatly increase the safety of the practitioners. These scenarios are fully supported by the TETRis (http://www.ponrec.it/opendata/risultati/ricerca-industriale/pon01_00451/) project. The main goal of TETRis is to enrich functionalities, services and applications of the TETRA communication system to foster the development of new open-source services for public safety scenarios. The reference architecture for TETRis is shown in Figure 1. At the lower layer, the networks – and TETRA in particular – are used to support the communication among the smart objects. A smart object is a physical/digital object, augmented with sensing, actuating, processing and networking capabilities. The smart objects layer coordinates and orchestrates the interaction among the smart objects to support the provisioning of value added services, such as environmental monitoring, emergency management, mobility control in urban areas, disasters prevention to the smart environments, namely “a physical world that is richly and invisibly interwoven with sensors, actuators, displays, and computational elements, embedded seamlessly in the everyday objects of our lives, and connected through a continuous network” (Posland, 2009). In this architecture, each WSN – and in particular the collection point (i.e., sink) of the sensor network – is

Extending TETRA with wireless sensor networks

5

interfaced with a smart object and provides real-time data employed to support the provisioning of advanced services for the smart environments. Figure 1

A reference architecture for TETRis networks and services (see online version for colours)

Figure 2

The MagoNode platform (see online version for colours)

2.1 TETRis application scenario We envision a scenario in which a rescue team is exploring an earthquake-stricken area. The team deploys one or more WSNs capable to collect real-time data that can prevent or limit future injuries or damages to the population and to the infrastructures. The WSN is able to monitor data related to structural health monitoring of damaged buildings as well as the quality of the air that could be impaired due to gas leakages. The network has to operate for weeks or months forwarding the collected data to a smart object that, using the TETRA network (due to the earthquake cellular networks are not operational), delivers the acquired information to a control and management point.

6

3

M. Paoli et al.

MagoNode: a new low-power WSNs platform

WSNs (see Akyildiz and Vuran, 2010; Stankovic, 2008) are ad hoc wireless networks made of battery powered electronic devices capable to coordinate their activities in order to collect real-time data on phenomena of interest. The limited energy availability provided by the batteries is the main bottleneck for the employment of WSNs in long-lasting monitoring tasks (Anastasi et al., 2009). As a consequence, a number of hardware and software solutions have been proposed in the literature to increase the energy efficiency of WSNs and thus prolong their network lifetime. These solutions include low-power hardware platforms (see Karani et al., 2011), low-power network protocols (see Demirkol et al., 2006; Akkaya and Younis, 2005), and the more recent energy harvesting techniques (see Seah et al., 2009), that try to collect energy from renewable sources (e.g., light, wind, etc.). In this section, we introduce the low-power MagoNode (see Paoli et al., 2013), a wireless mote developed at the Department of Computer, Automatic and Management Engineering (DIAG) of Sapienza – University of Rome. It features state-of-the-art low-power hardware and it is equipped with an 802.15.4 compliant transceiver operating in the ISM 2.4 GHz band. The 2.4 GHz band is available world-wide and provides a higher data rate when compared to other ISM bands. However, that band suffers from a lower transceiver sensibility and higher propagation losses that reduce the radio range of the motes. To improve the MagoNode radio range, we equipped the motes with a lowpower RF front-end that increases both the transmission power and the reception sensitivity, resulting in extended and reliable radio links at the cost of an energy overhead. Anyway, as discussed in Section 4.1, the high efficiency of nowadays transceivers, together with the low-power consumption of modern RF front-ends make this solution competitive in terms of energy efficiency with other widespread products available into the market. We stress that the reliability of radio links and the ability of covering large areas with a limited number of nodes, are crucial requirements for the effective adoption of WSNs for long-lasting monitoring activities in public safety scenarios. Notice that, while in the literature WSNs are usually assumed to be densely deployed, in real deployments the size of the network is limited by practical considerations such as: cost constraints, limited number of sensing points necessary to support effective monitoring (e.g., expert on structural monitoring use few sensing points per building) and limited accessibility to disaster areas.

3.1 Hardware architecture 3.1.1 Microcontroller and transceiver bundle The MagoNode platform is equipped with the Atmel Atmega 128 RFA1 (RFA1). It is a 8-bit, 16 MHz system-on-chip (SoC) 802.15.4 compliant with 128 KB of ROM, 16 KB of RAM, with outstanding transceiver performance (103.5 dB link budget, a current consumption of 12.5 mA in reception and 14.5 mA in transmission @+3.5 dBm) and supporting all the most common external hardware interfaces (I2C, UART, SPI). Being a microcontroller/transceiver bundle, the RFA1 chip helps in keeping the size of the PCB small and reduces the bill of materials with a significant improvement of the cost-effectiveness of the whole platform. Furthermore, the RFA1 is supported both by

Extending TETRA with wireless sensor networks

7

TinyOS (http://www.tinyos.net/) and by Contiki (http://www.contikios.org/), the two main operating system for WSNs.

3.1.2 RF front-end We equipped the MagoNode with an RF front-end to improve the reliability of wireless links and to extend the radio range of the motes in order to better cover wide areas with a limited number of nodes. An RF front-end is a bundle made of a power amplifier (PA) and a low noise amplifier (LNA) that is capable to both increase the transmission power and the reception sensitivity of the nodes (see Hao et al., 2010; Didioui et al., 2012), at the cost of an energy overhead that is discussed in Section 4.1. Nowadays, only few hardware platforms dedicated to WSNs feature an RF front-end; the most popular are: the AdvanticSYS CM3300, the Atmel ZigBit Amp ATZB-A24-UFL and the Dresden Elektronic deRFmega128-22M12. In Table 1, we compare the RF front-ends for WSNs available on the market. The selection of the most appropriate front-end should carefully consider two main criteria: 1

the limited energy consumption in receiving mode, since the idle listening is one of the major cause of energy waste in WSNs as pointed out in Demirkol et al. (2006)

2

local regulations for wireless devices, which are defined by organisations such as ETSI in Europe and FCC in North America.

Table 1 Brand

Front-ends comparison Model

Current RX (mA)

Gain RX (dB)

Max Pout (dBm)

Current TX (mA) @10 dBm

Current TX (mA) @20 dBm

Noise figure (dB)

Skyworks

SE2431L

5

12.5

23

50

115

2

Skyworks

SE2438T

5.5

10.5

16

20

N/A

3.5

Skyworks

SKY65344

7

10

20

N/A

105

2.2

Skyworks

SKY65352

7

10

20

72

110

2

RFMD

RF6555

8

11

20

60

90

3

RFMD

RF6575

8

13

22

90

170

2.5

Atmel

T7024

8

16

23

50

110

2.1

TI

CC2590

3.4

11.4

14

17

N/A

4.6

TI

CC2591

3.4

11

22

50

105

4.6

RFAxis

RFX2401C

10

12

22

N/A

100

2.4

RFAxis

RFX2411

8

12

21

N/A

110

2.5

In the specific, ETSI constrains the maximum output power (Pout) to +10 dBm (see ETSI, http://www.etsi.org/; Jennic, http://www.jennic.com/support/solutions/00002). Likewise, the FCC limits the maximum output power to +18–20 dBm (see FCC CFR 47, https://www.fcc.gov/; Jennic, http://www.jennic.com/support/solutions/00002). From Table 1, it can be noticed that the Texas Instruments CC2590 and CC2591 are the most energy efficient solutions (Current RX is 3.4 mA) that guarantee a satisfactory reception gain. As a consequence, according to the criteria discussed before, they have been

8

M. Paoli et al.

selected as the RF front-ends for the MagoNode respectively for European and North America deployments.

3.2 Software architecture One of the reasons for the selection of the Atmel ATmega1 28 RFA1 (RFA1) for the MagoNode lies in the fact that it benefits of a large software support. In particular, the RFA1 is supported by two of the most popular open-source operating system dedicated to WSNs and embedded systems: TinyOS and Contiki. Our current focus is on the creation of a stable and robust software platform for the MagoNode using TinyOS, anyway, we are also considering a porting on Contiki OS in the next future. TinyOS is an operating system dedicated to WSNs developed at Berkeley University. The open-source character of the project and its modularity are the main reasons of its success: the possibility to modify, reuse and adapt the source code to fulfil the developers need made TinyOS very popular and rich of applications and drivers for a great variety of hardware modules. Unlike traditional operating systems, designed to run on hardware architectures with large computational capabilities and without energy limits, TinyOS is designed in compliance with the typical constraints imposed by hardware modules for WSNs: low computational capabilities, limited memory capacity, limited power source.

3.2.1 Communication protocols TinyOS already supports a number of low-power communication protocols. In the implementation of our solution, we decided to exploit two well-known energy efficient protocols: the low power listening (LPL) and the collection tree protocol (CTP). The LPL is a duty-cycle MAC protocol (Moss et al., http://www.tinyos.net/tinyos-2.x/doc/html/ tep105.html), that increases the energy efficiency of the communications lowering the idle listening time of receivers, that as already observed, is a major component of energy expenditure in wireless communications. CTP (Gnawali et al., 2009; Colesanti and Santini, 2011) is a reliable and robust routing protocol for the collection of the data sensed by motes. CTP is a tree-based collection protocol: nodes in the network form a set of routing trees in order to reach nodes that advertise themselves as tree roots. The selection of the most ‘reliable’ route is made using the expected transmissions metric (ETX), used as routing gradient. There is a plethora of MAC and routing protocols dedicated to WSNs (see Demirkol et al., 2006; Akkaya and Younis, 2005). We selected LPL and CTP because they have been widely tested in various projects and represent a de facto standard for data collection in WSNs; their joint reliability is crucial for the employment in public safety scenarios.

3.3 Sensors for the TETRis application scenario The MagoNode is characterised by a modular architecture, therefore it can be easily integrated with expansion boards featuring the additional sensor capabilities necessary to support the TETRis application scenario, namely sensors to monitor critical infrastructures and the quality of the air.

Extending TETRA with wireless sensor networks

9

3.3.1 The MNA ambient board for air quality The MNA ambient board (Figure 5) is used to monitor the quality of the air and it is equipped with four sensors: Co2Meter COZIR CO2 sensor, Sensirion SHT75 temperature and humidity sensor, Sharp GP2Y1010AU0F dust sensor, Figaro TGS2442 CO sensor. All those sensors are characterised by limited power consumption, making them appropriate for the long-lasting deployments envisioned in the TETRis scenario. The COZIR is an ultra low-power (3.5 mW) digital CO2 sensor with a measurement range between 0 ppm and 2,000 ppm. It features a low noise measurement (< 10 ppm), low peak current (33 mA) and fast warm-up time (< 3 s). The Sensirion SHT75 is a digital temperature and relative humidity sensor. It provides a fully calibrated digital output with a precision ranging from 8 to 12 bits and is characterised by a 0.5 mA current consumption during measurements and only 0.3 uA in sleep mode. The Sharp GP2Y1010AU0F is an analogue dust sensor which detects the reflected light of dust in air by means of an infrared emitting diode and a phototransistor with a limited current consumption of 11 mA. The Figaro TGS2442 is a low-power, CO analogue sensor with a measurement range between 30 ppm and 1,000 ppm. It features low sensitivity to alcohol vapour and high sensitivity to carbon monoxide.

3.3.2 The MNA multisensor board for structural health monitoring The MNA multisensor board (Figure 5) is meant to interface a great variety of analogue sensors including those dedicated to structural health monitoring, like crackmeters, strain gauges and inclinometers. In our testbed, we used the D313 crackmeters and the VK40 vibration wire strain gauge by SISGEO (http://www.sisgeo.com/). In order to acquire data from analogue sensors the MNA multisensor board is equipped with a high precision, high data rate analogue to digital convert (ADC) by Texas Instruments, the ADS1256 and some additional instrumentation amplifiers, comparators and precision voltage references. The ADS1256 is a low-noise, 24-bit ADC converter with a maximum output data rate of 30kSPS and eight single-ended inputs or four differential inputs.

3.3.3 The MNA Temp&Hum We also implemented the MNA Temp&Hum Board (see Figure 6), a simple temperature and humidity sensor board developed mainly for prototyping and debugging purposes and to serve as relay node in the network.

4

Experimental results

In this section, we present the results of the experiments performed to: a

evaluate the energy efficiency of the MagoNode with respect to other solutions available into the market

b

test a WSN made of MagoNode nodes in a scenario with similar characteristics to the TETRis application scenario introduced in Section 2.

10

M. Paoli et al.

4.1 Energy efficiency of the MagoNode The energy efficiency of the MagoNode has been evaluated primarily in terms of current consumption. We compared the European version (equipped with the CC2590) and the US version (equipped with the CC2591) of our platform with amplified and unamplified motes available into the market. Clearly, the employment of an RF front-end implies that the current consumption in transmission of the MagoNode is higher than the one of unamplified motes. However, we claim that in a real multi-hop context, where energy consumption is driven by re-transmissions and hop counts, the benefits introduced by the RF front-end can mitigate the effects of the energy overhead of our platform. To support our hypothesis with experimental data, we compared the consumption on the same topology of a network made of MagoNodes (European version) and on a network made of IRIS motes (http://www.memsic.com/wireless-sensor-networks/XM2110CB) (a common unamplified platform).

4.1.1 Comparison with amplified and unamplified motes To assess the current consumption of the MagoNode platform, we ran two sets of experiments. We first compared both European and US versions of the MagoNode with respect to state-of-art amplified wireless nodes: the Atmel ZigBitAmpATZB-A24-UFL (ZigBit) and the Dresden Elektronic deRFmega 128-22M12 (deRFmega). Then, we compared the European version of the MagoNode to unamplified motes like TelosB, IRIS and the Atmel Atmega128RFA1-EK1. The Atmega128RFA1-EK1 features the same MCU/Transceiver bundle as the MagoNode, but it is not equipped with an RF front-end. We sampled the current consumption with a Rigol DM3068 precision multimeter with a period of 100 μsec and we calculated the average current consumption of each platform in reception (RX), transmission (TX) and idle listening (IDLE); data for the deRFMega platform have been derived from the datasheet. The results of the comparison of the MagoNode with amplified motes are shown in Figure 3. When the transmission power is 20 dBm the performance of the US version of the MagoNode outcomes those of deRFmega and Zigbit by 12% and 21% respectively. When the transmission power is limited to 10 dBm according to the ETSI regulation, the European Version of the MagoNode CC2590 has a current consumption than is nearly 50% less than to others. Moreover, the IDLE current consumption of both CC2590 and CC2591 is lower than both the deRFMega (by 9%) and Zigbit (by 32%). Similar results are obtained when the RX consumptions of the MagoNode and ZigBit platforms are compared. It was not possible to make a comparison with the deRFMega since the RX current consumption was not shown in its datasheet. However, the vast majority of WSNs platforms are unamplified, thus a natural question is: what are the differences in terms of energy efficiency between amplified and unamplified motes? To answer to this question, we compared the current consumption of the MagoNode with that of widespread unamplified motes and the results are shown in Figure 4. The presence of the CC2590 RF front-end is reflected in the higher current consumption in TX than those of IRIS@3 dBm (29%), TelosB@0 dBm (45%) and [email protected] dBm (90%). On the other side, current consumption on both RX and IDLE are surprisingly lower than IRIS (by 19% and 11% respectively) and TelosB (by 35% and 19% respectively). Of course, the MagoNode equipped with the CC2590 cannot beat the

Extending TETRA with wireless sensor networks

11

performance of the EK1 since the latter, features the same MCU/transceiver bundle as the MagoNode, but without the RF front-end. These results seem to support our assumption that the advantages in terms of link reliability and wider radio range justify the employment of the RF front-end. To further support this assumption, we run the experiments described in the next paragraph. Figure 3

Consumption comparison between the MagoNode and amplified motes

4.1.2 Comparison in a real testbed scenario The experiments discussed above show that the current consumption of the European version of the MagoNode in RX and IDLE are similar to the ones of unamplified nodes. This result suggests that in a real multi-hop contexts, where energy consumption is driven by re-transmissions due to link unreliability, the benefits introduced by the CC2590RFfront-end can mitigate theTXenergy overhead of our platform when compared to unamplified nodes. To support our hypothesis with experimental data, we compared the consumption of two networks with the same topology deployed in an indoor testbed scenario: the first one made of MagoNode (European version) and the second one made by IRIS motes (a common unamplified mote).

12

M. Paoli et al.

Figure 4

Consumption comparison between the MagoNode and unamplified motes

Figure 5

(a) The MNA ambient board and (b) the MNA MultiSensor Board inside a IP56 box (see online version for colours)

(a)

(b)

Extending TETRA with wireless sensor networks Figure 6

13

The MNA Temp&Hum Board (see online version for colours)

We deployed the two networks at the basement of our department, with the topology depicted in Figure 7. The location is characterised by concrete walls – 70 cm to 100 cm thick – and steel doors, which shrink the radio range of the nodes. Each network is made of 20 nodes in the ‘TETRis’ configuration, namely running LPL and CTP. Each experiment lasted 24 hours and was repeated in different days of the week to avoid possible bias of the results. Table 2 summarised the parameters used to setup the communication stack. Every node transmits the metric of interest every 5 minutes [i.e., with an inter-packet interval (IPI) of 300 s]. We compare the two networks using the following metrics: •

Data delivery ratio (DDR): it is given by data packets transmitted by each node over packets actually received by the sink.

•

Total ReTx: it is the total count of re-transmitted packets

•

Average time has lived (THL): it is the average hop count which packets travel through to reach the sink.

•

Total Fwd: it is the overall number of data packets that was forwarded over the whole multi-hop network.

•

Total CC [TX, RX, IDLE]: it is the overall current consumption of the network in mJ when motes are respectively in TX, RX and IDLE state.

•

Total TS [TX, RX, IDLE]: it is the overall time spent by motes in TX, RX and IDLE state.

Table 2

MagoNode vs. IRIS testbed parameters

Parameters Inter-packet interval

Value

Parameters

Value 500 ms

300s

LPL wakeup period

Battery

2 × 1.5 V

LPL Listen

10 ms

Packet size

103 bytes

LPL delay after receive

20 ms

24 h

RF channel

11

Run time Nodes number

20

14

M. Paoli et al.

As shown in Table 3, the DDR is 100% for both networks, but to achieve this result, the MagoNode network generates a significantly lower amount of traffic. As an example, number of retransmitted packets by MagoNode is 77% less than the IRIS one. This is a clear indication of an improved link reliability that is further confirmed by the reduction of the number of parent changes as well as by the lower values of the THL metric. The motivation for this performance lies in the improvements given by the RF front-end that brings to an overall reduction of the power consumption of the MagoNode network compared to the IRIS one. Table 3

MagoNode vs. IRIS testbed results [platform-name]

IRIS

Avg DDR (%)

100

100

0

Tot Parent Chgs

38

114

–66

Tot ReTx

319

1393

–77

Avg THL

1.55

2.14

–28

Tot Fwd Tot CC TX (mJ) Tot CC RX (mJ)

Δ (%)

3395

6804

–50

130379

132655

–1.7

37973

39391

–3.5

1713799

2260722

–24

Tot TS TX(s)

1569

2066

–24

Tot TS RX (s)

873

729

+2

33803

39662

–15

Tot CC IDLE (mJ)

Tot TS IDLE (s)

Note: Δs are calculated with respect to IRIS values. Figure 7

MagoNode vs. IRIS testbed area

Extending TETRA with wireless sensor networks

15

4.2 TETRis testbed The experiments discussed in the previous sections confirmed that the MagoNode is a convenient solution for the deployment of long-lasting WSNs. In this section, we discuss the results of a deployment in an hybrid indoor/outdoor testbed that is consistent with the reference TETRis scenario, presented in section 2. This testbed has been used as one of the final demonstrators of the TETRis project. An interview with engineers experts on structural health monitoring provided us ‘minimalistic’ requirements for the considered scenario: only few sensors per buildings are necessary and a sample per node every 15 minutes is considered sufficient to provide meaningful measurements. The testbed is made by 11 nodes placed at the basement and at the roof (see Figure 8) of the Department of Computer, Automatic and Management Engineering (DIAG) of Sapienza – University of Rome. All the protocols (i.e., LPL and CTP, see Section 3) were tuned to optimise the energy consumption, and consequently the network lifetime, in view of the sampling requirement. Figure 8

The TETRis testbed area

Indoor nodes are two MNA Ambient Boards, featuring CO, CO2 and dust sensors, and thus capable to collect a number of relevant data on the quality of the air in two adjacent rooms. Four MNA Temp&Hum boards (see Table 4) sampling temperature and humidity, are placed along the corridors. The structural health monitoring activity is performed by two MNA multisensor boards: one is placed indoor and it is interfaced to a vibrating wire strain gauge, while the other is placed outdoor and it is connected to a displacement sensor. Finally, two additional MNA Temp&Hum boards acting as relay nodes were deployed outdoor to manage wireless connectivity between the basement and the roof.

16

M. Paoli et al.

Table 4

TETRis testbed hardware composition

Id

Board

Sensors

Power

Note

0

MNA board

N/A

USB Port (5 V)

Sink, indoor

2

Ambient board

CO, CO2, Temp., Hum.

4xAA Alkaline

Indoor

3

MNA board

Temp., Hum.

4xAA Alkaline

Indoor

4

MNA board

N/A

2xAA Alkaline

Relay, outdoor

5

MNA board

Temp., Hum.

2xAA Alkaline

Indoor

6

MNA board

Temp., Hum.

2xAA Alkaline

Indoor

7

MNA board

Temp., Hum.

2xAA Alkaline

Indoor

9

MNA board

N/A

2xAA Alkaline

Relay, outdoor

11

Multisensor board

VW strain gauge

4xAA Alkaline

Indoor

15

Multisensor board

Displacement

4xAA Alkaline

Outdoor

19

Ambient board

Dust, CO2, Temp., Hum.

4xAA Alkaline

Indoor

Table 5

TETRis testbed parameters

Parameters

Value

Parameters

Value

Inter-packet interval

900 s

LPL wakeup period

500 ms

Packet size

50 bytes

LPL listen

5 ms

Run time

15 days

LPL delay after receive

20 ms

11

RF channel

16

Nodes number

The LPL wake-up period of each node is set to 500 ms and the listen time is set to 5 ms. This allows us to achieve a base duty cycle of the radio of 1%, which is sufficient to guarantee a network lifetime in the order of months. Notice that, the overall duty cycle is slightly larger than 1% (see Table 6) due to events that cannot be fully controlled, such as the traffic generate by the CTP routing layer and the retransmissions of data packets. The overall power consumption budget has also to take into account the heterogeneous consumption of the hardware interfaces described in Section 3.3. In particular, the CO and CO2 sensors are by far the most energy demanding. However, due to the relatively large sampling interval suggested by the domain experts, the energy necessary to sensing is marginal with respect to the energy employed in wireless communications. Finally, we enclosed the outdoor nodes into IP56 boxes, in order to protect the hardware from the agents of weathering. The impermeability of the casing is mandatory since the network will be deployed in harsh environmental condition. Table 5 summarises all the settings used in the testbed. The testbed started on December 2013 and ran one month. In this period, outdoor sensors have been exposed to eight days of rain with temperatures ranging between 2 to 15 degree Celsius. The results, shown in Table 6, point out the high reliability of the CTP that guaranteed a DDR above 99.5% for all the nodes of the network. High reliability is a crucial requirement in public safety scenario in order to elaborate trustworthy statistics on the state of the monitored area. As expected, the observed duty cycle ranges between 1.13% and 1.24% and it is thus slightly larger then the base one (1%). However, such increase does not impact significantly on the expected network life-time that is in the order of months.

Extending TETRA with wireless sensor networks Table 6 Id

17

TETRis testbed results DDR (%)

Duty cycle (%)

HopCount

Parent changes 0

0

100

100

0

2

100

1.13

1

0

3

100

1.21

1.93

4

4

99.65

1.22

2.73

3

5

100

1.2

1.68

4

6

100

1.2

2.02

4

7

100

1.17

2.06

4

9

99.86

1.19

3.73

1

11

100

1.18

1

0

15

99.86

1.21

4.58

6

19

100

1.24

1

0

Finally, the average network diameter, i.e., the average number of hops that a packet has to traverse to reach the sink, is only 2.2. This is a remarkable result, considering that the radio signal has to traverse concrete walls with thickness ranging from 70 cm to 100 cm, and a number of metal structures. Furthermore, the number of parent changes in the CTP protocol is significantly low; nodes change their parent due to bad links only in few occasions. This is an indirect evidence of a good network stability. The combination of small network diameter and good link stability is a direct consequence of the improved link quality provided by the RF front-end of the MagoNode platform.

5

Conclusions and future work

In this paper, we described the work done in the scope of the TETRis project, to design a WSN aimed to extend the functionalities of TETRA, in order to better support the management of critical situation with real-time feedback acquired on the field. We first introduced the MagoNode, a new mote family used in our testbed, detailing its hardware and software architecture and highlighting advantages and disadvantages. In particular, we carried out a current consumption comparison between the MagoNode and other platforms available on the market. The results showed that the MagoNode has lower current consumption in both transmission, reception and idle listening than other amplified motes. Moreover, while our platform has a higher current consumption in transmission than unamplified motes, in a real testbed scenario the advantages introduced by an RF front-end, i.e., extended radio ranges of the motes and more reliable links, allows the MagoNode to better suit many real application scenarios. Then, we showed how the MagoNode platform can be easily interfaced to a number of sensors and in particular to the sensors necessary for monitoring critical infrastructure and the air quality. Finally, the results of the experimental activity on a real testbed, confirm that WSNs can be effectively used to support the management of critical situations in the considered scenario and that the MagoNode platform well meets the requirements provided by experts on structural health monitoring. In the future, we plan to extend the

18

M. Paoli et al.

testbed considering a higher number of nodes and interfacing other sensors boards designed to support specific applications.

References Akkaya, K. and Younis, M. (2005) ‘A survey on routing protocols for wireless sensor networks’, Ad Hoc Networks, Vol. 3, No. 3. Akyildiz, I.F. and Vuran, M.C. (2010) Wireless Sensor Networks, Vol. 4, John Wiley & Sons. Anastasi, G., Conti, M., Di Francesco, M. and Passarella (2009) ‘A energy conservation in wireless sensor networks: a survey’, Ad Hoc Networks, Vol. 7, No. 3, pp.537–568, ISSN: 1570-8705. Chintalapudi, K., Fu, T., Jeongyeup, P., Kothari, N., Rangwala, S., Caffrey, J., Govindan, R., Johnson, E. and Masri, S. (2006) ‘Monitoring civil structures with a wireless sensor network’, Internet Computing, IEEE. Colesanti, U.M. and Santini, S. (2011) The Collection Tree Protocol for the Castalia Wireless Sensor Networks Simulator, Technical Report Nr. 729, ETH Zurich. Contiki [online] http://www.contiki-os.org/. Demirkol, I., Ersoy, C. and Alagöz, F. (2006) ‘MAC protocols for wireless sensor networks: a survey’, Communications Magazine, IEEE, Vol. 44, No. 4. Didioui, A., Bernier, C., Morche, D. and Sentieys, O. (2012) ‘Impact of RF front-end nonlinearity on WSN communications’, ISWCS 2012. Digital Mobile Radio (DMR) [online] http://www.etsi.org/technologies-clusters/technologies/ digitalmobile-radio. European Telecommunications Standards Institute (ETSI) [online] http://www.etsi.org/. FCC CFR 47 [online] https://www.fcc.gov/. Gnawali, O., Fonseca, R., Jamieson, K., Moss, D. and Levis, P. (2009) ‘Collection tree protocol’, SenSys’09. Gomez, L., Laube, A. and Ulmer, C. (2009) ‘Secure sensor networks for public safety command and control system’, Technologies for Homeland Security, HST ‘09. Hao, Z., Zhiqun, L., Meng, Z. and Gang, C. (2010) ‘A2.4GHz low-IF RF frontend for wireless sensor networks’, ICMMT 2010. Hill, J., Horton, M., Kling, R. and Krishnamurthy, L. (2004) ‘The platforms enabling wireless sensor networks’, Communications of the ACM, Vol. 47, No. 6, pp.41–46. IRIS mote [online] http://www.memsic.com/wireless-sensor-networks/XM2110CB. Jennic [online] http://www.jennic.com/support/solutions/00002. Karani, M., Kale, A. and Kopekar, A. (2011) ‘Wireless sensor network hardware platforms and multi-channel communication protocols: a survey’, IJCA Proceedings on 2nd National Conference on Information and Communication Technology. Kortuem, G., Kawsar, F., Fitton, D. and Sundramoorthy, V. (2010) ‘Smart objects as building blocks for the Internet of things’, Internet Computing, IEEE. Mainwaring, A., Culler, D., Polastre, J., Szewczyk, R. and Anderson, J. (2002) ‘Wireless sensor networks for habitat monitoring’, in Proceedings of the 1st ACM International Workshop on Wireless Sensor Networks and Applications, ACM, pp.88–97. Moss, D., Hui, J. and Klues, K. (xxxx) Low Power Listening – TEP105 [online] http://www.tinyos.net/tinyos-2.x/doc/html/tep105.html. Paoli, M., Lo Russo, A., Colesanti, U.M. and Vitaletti, A. (2013) ‘MagoNode: advantages of RF front-ends in wireless sensor networks’, RealWSN 2013. Posland, S. (2009) Ubiquitous Computing Smart Devices, Smart Environments and Smart Interaction, Wiley.

Comment [A4]: Author: Please provide the page numbers. Comment [A5]: Author: Please provide the place of publication.

Comment [A6]: Author: Please cite the reference in the text or delete from the list if not required. Comment [A7]: Author: Please provide the access details (date when the site was accessed/visited).

Comment [A8]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A9]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A10]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A11]: Author: Please cite the reference in the text or delete from the list if not required. Comment [A12]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A13]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A14]: Author: Please cite the reference in the text or delete from the list if not required.

Comment [A15]: Author: Please provide the year of publication and access details (date when the site was accessed/visited). Comment [A16]: Author: Please provide the place of publication.

Extending TETRA with wireless sensor networks

19

SAFECOM Program (2004) Statement of Requirements for Public Safety Wireless Communications and Interoperability. Seah, W.K.G., Eu, Z.A. and Tan, H-P. (2009) ‘Wireless sensor networks powered by ambient energy harvesting (WSN-HEAP): survey and challenges’, Wireless VITAE. Sha, K., Shi, W. and Watkins, O. (2006) ‘Using wireless sensor networks for fire rescue applications: requirements and challenges’, IEEE International Conference on Electro/information Technology. SISGEO [online] http://www.sisgeo.com/. Stankovic, J.A. (2008) ‘Wireless sensor networks’, IEEE Computer, Vol. 41, No. 10, pp.92–95. Terrestrial Trunked Radio (TETRA) [online] http://www.etsi.org/technologiesclusters/technologies/tetra. TETRA Critical Communication Association (TCCA) [online] http://www.tandcca.com/. TETRA vs. DMR (2012) [online] http://www.tandcca.com/Library/Documents/TETRAversusDMRsmeOct2012.pdf. TETRA Innovative Open Source Services (TETRis) project [online] http://www.ponrec.it/opendata/risultati/ricerca-industriale/pon01_00451/. TinyOS [online] http://www.tinyos.net/.

Comment [A17]: Author: Please cite the reference in the text or delete from the list if not required.

Comment [A18]: Author: Please cite the reference in the text or delete from the list if not required. Comment [A19]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A20]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A21]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A22]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A23]: Author: Please provide the access details (date when the site was accessed/visited). Comment [A24]: Author: Please provide the access details (date when the site was accessed/visited).