Impact Evaluation of Radio over Fiber Technology in Wireless Sensor Networks

Transcription

1 Impact Evaluation of Radio over Fiber Technology in Wireless Sensor Networks Raphael M. Assumpção, O. C. Branquinho and M. L. F. Abbade School of Electrical Engineering Pontifical Catholic University of Campinas D. G. Lona School of Electrical and Computer Engineering State University of Campinas Arismar Cerqueira S. Jr School of Technology State University of Campinas Abstract We analyze the impact of a Radio over Fiber technology in a Wireless Sensor Network based on FSK and MSK modulation schemes. Experiments were carried in a WSN testbed at 915 MHz frequency range to verify the impact of RoF used as a backhaul in the sensor coverage. Moreover, it has been investigated the impact of the introduction of multiple cells. In order to avoid radio propagation uncertainties a channel emulation system was applied and the performance parameters measured were Packet Error Rate and Received Signal Strength Indicator. The results demonstrate the viability of the use of RoF as a backhaul for WSN, considering the correct impact in the coverage distance by the use of an appropriate low noise amplifier. Keywords-Wireless Sensor Network; Radio over Fiber; noise; coverage I. INTRODUCTION Wireless Sensor Network (WSN) technology has become a reality due to the recent advances of microelectronics [1] [2]. It has been used in many areas, such as petroleum industry [3], coal mines [4] and even agro-climatology [5] for the control and monitoring of different physical properties, such as temperature, pressure, humidity, toxic gases, and others. Radio over Fiber (RoF) systems are used to transport the Radio Frequency (RF) signal through optical fiber links. This is accomplished by analogically modulating a laser with a radio frequency (RF) signal [6]. RoF systems can provide specialized coverage of wireless services by using an extended optical backhaul. This technical strategy makes possible to have remote antenna units (RAU) deployed in distant areas and have all your base station (BS) equipment centralized [7]. These systems are suitable for variety applications, such as Inbuilding coverage, outdoor cellular systems and broadband fixed and mobile wireless access. With appropriate technical considerations they are entirely transparent to the system frequency, protocol and bit rate [6]. This characteristic makes them extremely interesting for the convergence of optical and mobile systems. Although there are many papers in literature regarding WSN and RoF technologies, just a few of them report their implementation in an unique technological solution [8] [9] [10]. We have recently reported the implementation of a RoF system based on IEEE standard and using Thin architecture in a geographically-distributed optical network [8]. Where simulation and experimental results have shown no signal degradation and demonstrated RoF technology can be very interesting as a backhaul for WSNs with reach of up to 600 km. The only penalization to be considered is the noise figure as demonstrated in the next sections. In this work, we study the possible sources of RF signal degradation in the RoF systems and verified the impact of using RoF as a backhaul to WSN. The impact in radio coverage is analyzed, as well as the impact of deploying multiple RAUs RoF system using low cost commercial RoF equipment. To the best of our knowledge, this is the first time that such analysis is presented in literature. The tests were performed with FSK and MSK modulation, using a data rate of 250 Kbps. The results were obtained changing the channel attenuation while the PER (Packet Error Rate) and RSSI (Radio Signal Strength Indicator) were evaluated. The tests permitted the evaluation of the effect of RoF in the WSN coverage. The results identify the necessary adjusts in the RoF and WSN system parameters to obtain the desired coverage in the planning of such system. The remaining of this work is organized in following way. In Section II, we review some of the fundamental aspects of RoF technology that affect the WSN signal. In Section III, we describe the experimental setup utilized in our experiments. This work is part of FOTONICOM Program, supported by FAPESP (grant 08/ ) and by CNPq (grant /2008-9). It was also supported by CNPq under grant /

2 Results are presented in Section IV. Finally, our conclusions are approached in Section V. II. RADIO OVER FIBER SYSTEMS Radio over Fiber systems transmits RF signal through an optical fiber by analogically modulating a laser to convert this signal from the electrical domain to the optical domain. This optical signal can be generated by direct or external laser modulation. External modulation is usually carried out with Mach-Zehnder interferometer (MZI). But direct modulation allows to modulate a signal with a few GHz and still not use a MZI, which leads to lower cost products [10]. The RoF equipment used in this study uses direct modulation. For a direct modulation system the gain can be defined as shown in (1) [11]. G ( f ) d = Sl α Sd ( f ) (1) Rl where G is the gain of the RoF in the RF domain, S l (f) is the laser slope efficiency, α is the optical attenuation, S d (f) is the detector slope efficiency, R d is the detector load resistance and R l is the effective laser resistance. Equation (1) indicates that the optical attenuation impact in the gain is squared. The Noise Figure (NF) of non-amplified RoF system comes from noise contributions originated in the light generation and in the photo detection processes. The most important ones are laser relative intensity noise (RIN), thermal noise, and shot noise. RoF NF can be estimated by (2) [12]. 2 I D RIN Rd 2q I D Rd 1 NF = 10 log (2) k T G k T G G where, I D is the current at the photo detector output, RIN is the laser relative intensity noise, R d is the detector load resistance, k is the Boltzmann constant, T is the temperature in K, G is the RoF gain and q is the charge of an electron. In commercial systems the sum from all the noise contributions is presented as equivalent input noise (EIN) [13] given by: EIN = R + R + R (3) RIN SHOT T R addition of a new cell. This analysis is case specific since the noise which limits each system can vary. III. EXPERIMENTAL SETUP For testing the WSN degradation imposed by the RoF a channel emulation bench was assembled using a variable attenuator to emulate the channel attenuation. This attenuator emulates the sensor being carried away from the antenna located in the RoF toward the cell boundaries. In order to evaluate the RoF impact on WSN communication packet error rate (PER) and received signal strength indicator (RSSI) information were collected throughout the tests. Each experiment consisted of a series of tests where the attenuation was slowly raised while PER and RSSI date where collected. For each point 5 measures of 10 thousand packets were transmitted from the sensor to the base station. We used two different modulations, FSK and MSK, in the radio sensor during the tests to verify if both had the same behavior. In both cases the transmission rate was set to 250 kbps. In the FSK modulation a filter bandwidth of khz was used. For the MSK modulation the filter bandwidth was khz. Packets used for the transmission were 152 bits long. Fig. 1 presents all the packet fields and their corresponding bit lengths. Preamble Sync word Length Destination address Source address Payload CRC 32 bits 32 bits 8 bits 8 bits 8 bits 48 bits 16 bits Figure 1. Packet format Fig. 2 shows the WSN with RoF testbed layout. It is modular and can be reorganized to match the proposed tests. Where R RIN is the RIN noise, R SHOT the shot noise and R T the thermal noise. EIN is usually presented in dbm/hz units. Considering that the addition of another laser in the fiber will add another RIN noise contribution to the system and also add more optical power that is translated as shot noise in the photodetection. we can then conclude that the EIN for a system with more cells is given by: EIN = R + R + R (4) RIN SHOT Where the summation represents the sum of all RIN noises. Direct modulated RoF systems are RIN limited systems for small optical attenuation values [14]. By this we concluded that in our system where both equipments are similar and have the same attenuation, that our noise will be almost doubled by the T Figure 2. Testbed layout The testbed is composed of a sensor, a base station, a variable attenuator to emulate the radio channel, an adjustable power supply to control the attenuator, antenna duplexers, 3 Identify applicable sponsor/s here. If no sponsors, delete this text box. (sponsors) 619

3 commercial RoF equipments, 2 optical couplers, a sensor working as base station and a computer for data processing and logging. The commercial RoF system used have a wavelength in the range of 1310 nm, the optical output is 3dBm, and typical RF gain without optical attenuation of 0 db. The EIN presented in the datasheet is -134 dbm/hz and the RIN value was given to us by the manufacturer and is -140 db/hz Our testbed have an optical attenuation of 6.5 db between RoF Cells and RoF BS. This was reflected as a measured RF gain of -15 db in our tests. In the first experiment the base station was connected right after the variable attenuator without any RoF. The results from this experiment were used as a reference for the other tests. In the second experiment the RoF from the BS and the first cell (cell 1) were introduced in the test. No LNA was connected but the optical couplers were used in order to make fair the comparison of the results with one and two cells. In the third experiment the other RoF (cell 2) was connected to the testbed. Since we just wanted to measure the degradation caused by this other cell, no sensor was connected to it (i.e., the RoF equipment on cell 2 emitted a continuouswave signal). After those three tests were made for both modulations, FSK and MSK, they were repeated using the LNA for expanding the cell coverage distance. Using the results from our experiments we estimated the coverage reach by using a log-distance path loss model [15] considering also the LNA and RoF gains as presented in (5). 0 P + G TX TX + G RX 4πd0 10log + G λ 10β d = d *10 (5) Where d is the coverage radius, d 0 is reference distance used in the log distance model, P TX is the RF transmission power, G TX and G RX are, respectively, the transmitter and receiver antenna gains, λ is the RF wavelength, G RoF is the RoF gain, G LNA is the gain from de LNA, P RX is the received power needed for each of the tests and β is the loss exponent used in the path loss model. In our tests d 0 = 1 m, P TX = 10 dbm, G TX = G RX = 0 db, λ= m, G ROF = -15 db, G LNA = 13.5 db, β=3.41. The utilized value of β stands for an open area for WSN [16]. In cases where RoF and/or LNA were not used the gain was considered 0 db. 2 RoF + G IV. RESULTS Our experiments results are PER x RSSI measured values. We considered that the maximum tolerated PER would be 5%, which is high for most data network systems but can be considered acceptable for some WSN applications [17]. The results from our experiments are presented grouped by different modulation and by the use of LNA. Firstly it will be presented the results from the tests without LNA, with FSK in LNA P RX Fig. 3 and MSK in Fig.4. Next we will present the results with LNA, FSK in Fig.5 and MSK in Fig. 6. The last results are the coverage analysis. Figure 3. Results for FSK Without LNA Fig. 3 presents the results for the FSK modulation. For the WSN only, our reference test, we have a PER of 5% with a RSSI of -95 dbm. Using the RoF as a backhaul the same PER of 5% was reached with a RSSI of -71 dbm. For the test with 2 covering cells the 5% PER was reached with a RSSI of dbm. These results represent a degradation of 24 db for the use of one RoF and 26.4 db for 2 RoF cells. Figure 4. Results for MSK Fig. 4 presents the results for MSK modulation. The performed tests for 5% PER lead to an RSSI of -96, -72.7, and dbm, respectively, for the cases where no ROF, a single RoF and two RoFs were considered. This stands for a degradation of 23.3dB and 26.3 db for the cases of a single RoF and two RoFs. It is interesting to note that the MSK did not have the expected performance of 5 db better than FSK [15], in the test the difference was approximately of 1 db. We concluded that the transceiver implementation did not have the best performance with this modulation. 620

4 Figure 7. FSK coverage radius Figure 5. Results for FSK with LNA Fig.7 presents the coverage for the FSK modulation estimated for a sensor transmitting with 10 dbm while 0 dbi antennas were used. Fig. 5 presents the results when the LNA was used with the FSK modulation. The test with no RoF resulted in a 5% PER with a RSSI of dbm. With the LNA and 1 RoF covering cell the RSSI was dbm while for 2 RoF covering cells it was dbm. The inclusion of the LNA increased the coverage distance. However for the necessary RSSI for the 5% PER was greater 4.5 db. This is not a degradation because the coverage radius was greater. This value is the LNA noise figure measured by the receiver. For 1 RoF with LNA the degradation was 23.1 db and for 2 RoF cells with LNA it was 27.7 db. Figure 8. MSK coverage radius Fig.8 presents the estimated coverage for the MSK modulation. From the estimated reaches obtained is possible to observe that the RoF introduced a huge degradation in the RF coverage. it is also possible to observe that the LNA is able to counteract some of the degradation caused by the RoF, thereby improving the coverage of the WSN. From these results it is possible to infer that most of the degradation observed is caused by the noise added by the RoF. Figure 6. Results for MSK with LNA Fig. 6 presents the results when the LNA was used with the MSK modulation. The test with no RoF resulted in a 5% PER with a RSSI of -91,4 dbm. With the LNA and 1 RoF cell the RSSI was dbm. The RSSI for 5% PER with 2 RoF cells was dbm. For the MSK modulation the same effects were observed as in FSK. Results show that the behavior of the ROF system does not depend on the modulation scheme used in the RF interface. The RSSI results obtained in our tests were then used in (5) as P RX value to determinate the covering reach for each case tested. V. CONCLUSIONS In this work we presented an experimental investigation of the performance of RoF technology applied as backhaul to WSNs. FSK and MSK modulation schemes were considered in scenarios with and without the use of LNA. Results indicated that inclusion of RoF equipment introduces appreciable noise degradation to the system and severely deteriorates the coverage of WSN RF cells. However, such deleterious effect may be easily overcome by using commercial LNAs. So the results demonstrate that it is possible to use the RoF as backhaul to WSN, but it is necessary a carefully analyzes of the noise figure introduced by the RoF to obtain an adequate WSN cell coverage. Future works should focus on WSN MAC protocols for optimizing WSN with ROF Backhaul systems. ACKNOWLEDGMENT The authors wish to acknowledge the financial support from FINEP in the research laboratory LP-SiRa of PUC- Campinas. 621

5 REFERENCES [1] I. F. Akyildiz, et al., "Wireless sensor networks: a survey," Computer Networks, vol. 38, pp , [2] C. Y. Chong and S. P. Kumar, "Sensor networks: Evolution, opportunities, and challenges," Proceedings of the Ieee, vol. 91, pp , Aug [3] A. J. T. S. Mello, et al., "Uma Rede de Sensores para Monitoração da Proteção Catódica em Dutos," presented at the 4o PDPETRO, Campinas, [4] M. Li and Y. Liu, "Underground coal mine monitoring with wireless sensor networks," ACM Trans. Sen. Netw., vol. 5, pp. 1-29, [5] T. M. Johnson and M. Margalho, "Wireless Sensor Networks for Agroclimatology Monitoring in the Brazilian Amazon," in Communication Technology, ICCT '06. International Conference on, 2006, pp [6] H. Al-Raweshidy and S. komaki, Radio over Fiber Technologies for Mobile Communication Networks. Norwood, MA: Artech House, [7] P. H. Gomes, et al., "Optimization of the Use of Radio Resource of Radio-Over-Fiber Access Networks," in GLOBECOM 2010, 2010 IEEE Global Telecommunications Conference, 2010, pp [8] Arismar Cerqueira S. Jr., et al., "Performance analysis of a Radio over Fiber system based on IEEE standard in a real optical network," Microwave and Optical Technology Letters, vol. 51, pp , [9] T. Jun, et al., "A Hybrid Radio Over Fiber Wireless Sensor Network Architecture," in Wireless Communications, Networking and Mobile Computing, WiCom International Conference on, 2007, pp [10] M. Hossen, et al., "Extension of wireless sensor network by employing RoF-based 4G network," in 11th International Conference on Advanced Communication Technology, 2009, pp [11] A. Wiberg, "Generation, Modulation and Detection of Signals in Microwave Photonic Systems," PHD, Department of Microtechnology and Nanoscience, Chalmers University of Technology, Göteborg, [12] E. I. Ackerman and C. H. Cox, III, "State of the art in analog fiberoptic link technology," in Signals, Systems, and Electronics, ISSSE URSI International Symposium on, 1998, pp [13] X. N. Fernando, "Signal Processing for Optical Fiber Based Wireless Access," Doctor of Philosophy, Department Of Electrical And Computer Engineering, University of Calgary, Calgary, [14] X. N. Fernando and A. Anpalagan, "On the design of optical fiber based wireless access systems," in Communications, 2004 IEEE International Conference on, 2004, pp Vol.6. [15] T. S. Rappaport, Wireless communications : principles and practice, 2nd ed. Upper Saddle River, N.J.: Prentice Hall PTR, [16] A. TIWARI and S. S. THAKUR, "Consequence of Urbanization on Wireless Sensor Network Reportage Domain," International Journal of Engineering Science and Technology, vol. 3, [17] H. Karl and A. Willig, Protocols and architectures for wireless sensor networks. Hoboken, NJ: Wiley,