Simulation of wireless channel for UAV communication

Transcription

1 1 Simulation of wireless channel for UAV communication Tiago C. S. Xavier, Elvio J. Leonardo, João A. Martini, Itana M. S. Gimenes and Luciana A. F. Martimiano Abstract Unmanned Aerial Vehicles (UAVs) have been used intensely in various activities such as surveillance of geographical boundaries, monitoring of agricultural areas and image capturing. Most systems for these types of applications require some way of communication between the aircraft and stations placed on the ground and usually they have severe limitations regarding energy availability. This makes the use of technologies that promote energy efficiency very important. Thus, designing communications systems for UAVs is a challenging topic and one that has attracted much interest. This paper analyzes a communication channel for UAVs in the presence of noise, Doppler shift and fading. Moreover, simulations are presented to estimate the bit error rate and the coverage area The simulations were performed using typical parameters for the ZigBee protocol since it appears as a viable solution communication problem presented. Index Terms Coverage area, Rice, UAV, ZigBee T I. INTRODUCTION HE use of large aircrafts in certain activities such as monitoring large areas, collecting images, surveillance of large territories and others has a cost that is usually very high, and prevents those with a lower purchasing power to make use of this type of service. An alternative to this situation, which is currently the target of intense research, is the use of unmanned aerial vehicles (UAVs). The UAVs are small aircraft without crew, which are controlled remotely and have the ability to make autonomous flights. As examples, [1] and [2] show UAV developed primarily for use in agriculture and [3] illustrates a UAV Project to help on defense and security of air bases. In UAV systems a common situation is the need of communication between aircraft and ground transceivers, stations or computing devices that need to send and receive information for the aircraft in real-time. In this context, the design phase of the UAV communication system requires the Manuscript received February 14, This work was supported in part by the CNPq-INCT-SEC. TiagoC. S. Xavier is a student at the State University of Maringa; tiago.cariolano@ gmail.com. Elvio J. Leonardo ( ejleonardo@ uem.br), João A. Martini ( jangelo@ din.uem.br), Itana M. S. Gimenes (itanagimenes@gmail.com) and Luciana A. F. Martimiano (lmartimiano@gmail.com) are with the State University of Maringa.. analysis of the communication channel to determine what equipment and technologies should be used. The analysis of the communication channel has the goal to establish performance figures for the application under development. Among the key measures of performance of mobile communication links, two of major interest are: Bit Error Rate (BER) and coverage area. The first indicates what percentage of bits coming out of the transmitter arrive with an incorrect value at the receiver; and second determines the probability that the receiver is in a location where signal strength is satisfactory. To estimate these measures it is important a thorough analysis of the communication channel and what types of impairments can affect the signal. In the case of communication systems for UAVs, the fading, the frequency shift and the noise are of great relevance. Thus, in order to estimate the BER and coverage area it is necessary to take them into account. Another problem that also influences the communication systems for UAVs is the fact that the equipment requires low power consumption. Thus, it is necessary to use circuits, devices, protocols and communication technologies that take this aspect into consideration. To try to improve these types of systems, this paper analyzes a communication channel between a UAV and a ground station and it estimates what is the probability that the data reach the receiver incorrectly and the coverage area of the signal for the environment studied. The channel, due to the characteristics of the environment in which the UAV is inserted, is usually modeled in a way that the signal envelope follows the Rice distribution, with attenuation or apparent frequency shift due to the Doppler effect and Additive White Gaussian Noise (AWGN). Furthermore, the modulation schemes used are the Offset Quadrature Phase Shift Keiyng (O-QPSK) and the Binary Phase Shift Keying (BPSK). By modeling the communication channel in this way, the work presented here brings a contribution to the estimation of the BER and the coverage area for a communication channel with all these features together. In addition, in the simulations it is assumed that the equipment and devices support ZigBee s physical layer protocol [4, 10], which has among main features the low power consumption, that is a requirement for a UAV system.

2 2 The methodology applied in this work was the use of simulation employing the computational tool Simulink from Matlab [5]. The type of aircraft considered was the described in the reference [11]. This paper is organized as follows. Section 2 presents the analytical model; Section 3 shows the numerical results and the Section 4 presents the conclusions and proposals for future work. II. ANALYTICAL MODEL Communication signals may suffer several alterations in its path from the transmitting antenna until it reaches the receiving antenna such as diffraction, scattering, absorption and reflection. In this work, it is considered relevant changes in the specific case of UAV system those caused by: noise, fading and apparent frequency shift. A. Noise The noise happens when the data signal is affected by others random signals, that occurs for various reasons such as wirelless phones networks, satellite signals and communication signals broadcast. For the work presented here, this type of noise is modeled as AWGN, which is commonly used in the analysis of wireless communication channels. Therefore, the amplitude of the noise follows the Normal (Gaussian) distribution with zero mean and probability density function (PDF) expressed by (1) where is its effective (or rms) value [6]. B. Fading The fading is caused by the multipath signal propagation. This phenomenon happens because the environment in which the aircraft flies is composed of several obstacles, both natural and artificial, which are located randomly on the ground. Rivers, mountains and buildings are examples of elements that can alter the signal physical characteristics, imposing different attenuations and phase changes. This random features of the environment makes it necessary to model the channel using a statistical approach. Over time a large number of distributions were proposed for various specifics cases. The environment of the UAV has a feature that is the aircraft most of the time have a line of sight with the ground station with which it communicates. This allows the fading, in this case, to be modeled by the Rice probability distribution, where the PDF of the signal envelope is expressed by (2) where is the modified Bessel function of order 0 [7, Eq ]. The Rice distribution has a parameter that determines the relationship between the powers of the direct and diffuse components of the signal, which is defined as [8]. (3) The higher the value of, the more dominant is the direct component in relation to the diffuse component. C. Doppler Effect Another phenomenon that occurs in the UAV communication system signal is an apparent carrier frequency shift, which is caused by the Doppler effect. This shift occurs by the relative motion between the transmitter and receiver, which happens when the aircraft is approaching or flying away from the ground transceivers and it is proportional to the speed of mobile. The frequency shift is expressed by (4) where is the mobile speed, is the wavelength of the carrier and is the direction of displacement of the mobile. D. Bit Error Rate The Bit Error Rate (BER) is a measure that determines what percentage of bits that are received incorrectly by the receiver. Of course, the lower the BER is the better a communication system is. Several factors such as fading, attenuation and frequency shift may cause the BER to have a high value, but, mostly, the main factor that affects it is the noise. Thus, the Signal to Noise Ratio (SNR) is a parameter widely used because of its usually close relation to the BER. E. Coverage Area The coverage area of the signal defines the geographical area where a mobile can communicate with a transceiver on the ground with a signal intensity. The approach used here is to determine the proportion of locations at a distance from the ground station where the received signal power is above a threshold power. This proportion is set on the perimeter of the circle defined by the radius. Of course, means that there is a 100% probability that the mobile receives a signal equal or larger than the threshold; and means that there is a 100% probability that the mobile to be outside the limits of cell coverage area and receives a signal lower than the threshold. F. Modulation The ZigBee s physical layer protocol has two modulation schemes that are used depending on the value of the carrier

3 3 frequency. For the frequency to equal 2.4 GHz, the modulation used is O-QPSK; and for frequencies of 868MHz and 902MHz, it is used BPSK [9]. Thus, the work presented here uses these two modulation schemes to analyze the communication channel. III. RESULTS A. Bit Error Rate Fig. 1 shows the BER as a function of SNR using O-QPSK modulation. For this simulation the aircraft speed was considered 100 km/h and the carrier frequency 2.4 GHz, so from equation (4) the maximum Doppler shift ( ) is calculated to. Moreover, to see what the diffuse components to dominate over the direct signal, which makes the influence of fading greater than the SNR. That is the reason why when, the influence of noise becomes smaller and the BER values are nearly the same, all very close of 0.5. In order to show the difference between the modulation schemes of the ZigBee protocol, the simulation of Fig. 2 shows the behavior of the BER with BPSK modulation. Since Fig. 2. BER for BPSK and = Hz. Fig. 1. BER for O-QPSK and = Hz. influence the fading intensity has on the BER, each line on the graph corresponds to a different value of Rice distribution s parameter. It can be seen from Fig. 1 that, in general, the BER is inversely proportional to the value of SNR and. This is expected, since as the value increases, the more the direct component overlaps the diffuse components; and a higher SNR translates to a lower probability that the receiver gets a incorrect bit. Also it can be seen that with and as grows the SNR value, the BER decreases rapidly with higher values of SNR we have BER <. This means that in this situation the aircraft or the ground stations are receiving a satisfactory signal. Still in Fig. 1, with smaller values of the BER tends to stabilize and it is not influenced by the variation of SNR. This case illustrates situations where the fading is very intense, as when the aircraft flies over areas with many obstacles, so the parameter is very low causing the carrier frequency used for this modulation is 868 MHz, the Doppler shift is set to. The others parameters remain the same. The plot of Fig. 2 has the same behavior as the previous Fig., but it can be noted that independently of or SNR values, all BER figures are better (smaller) when compared with the O-QPSK modulation. As an example, for all tested values of, BER < 0.5. B. Coverage Area Fig. 3 shows results for the coverage area. In the Fig., the abscissa corresponds to the power threshold and the ordinate is the probability, which is the proportion of locations at a distance L from a ground station where a mobile would experienced the received signal above the. In the simulation it was used the O-QPSK modulation scheme. The UAV speed was set to 100 km/h, the carrier frequency to 2.4 GHz, the maximum Doppler shift to and the SNR to 20 db. As in the BER estimation, the main parameter analyzed is the Rice distribution s parameter, so every curve of the plot represents the coverage area value for a different value of. In examining Fig. 3, when tends to minus infinity (the lowest ) we see that independently of parameter factor, all

4 4 the curves has. In contrast, as the value of increases, all the lines tend to. This behavior is expected, because indicates that a receiver is more likely to be within the coverage area if your reception sensitivity is higher, i.e., the lower the threshold, which means that the receiver is able to detect weaker signals, translates in a increase of the higher relative amount of noise, it can be noticed that there is a narrowing of the curves and generally they are all flattened down, which causes a greater degree of uncertainty of the coverage area, in addition to its reduction. If the two previous Fig. 3. Coverage Area for = Hz and SNR = 20 db. probability of the UAV to be within coverage area from transmitter. Still in the same Fig., it can be seen that for, the parameter factor is directly proportional to the coverage area. However for, approximately, the parameter factor is inversely proportional to the coverage area. This behavior occurs because the higher the value of, the more deterministic environment becomes, and thus the curve of coverage area tends to fall more and more perpendicularly indicating a sharper defined limit. For instance, in Fig. 3 we see that when the coverage area can be estimated more accurately, because for and the values of are equal 1 or 0 indicating that the signal is being received or not and only in the range the coverage area is in the range between 0 and 1. A very different situation occurs for, because regardless of value assigned to, the coverage area is fuzzier due to random influence of the diffuse components. In addition, when, whenever there is an increase in, meaning that there is an increase in the dominant component power in relation to the diffuse components, the coverage area is also larger; and for the increase of decreases the coverage area. Fig. 4 shows results for a similar simulation, except that. This situation indicates a deterioration in signal quality compared to the previous scenario. As a result of a Fig. 4. Coverage area for and. graphs are compared, it can be noticed that for any value of, the values of are no-deterministic (0 or 1), unlike what occurred when SNR = 20 db. Still, all the curves have an abrupt decrease from tends to minus infinity, which makes Fig. 5. Coverage Area for SNR = 20 db and k = 20 db. the coverage area considerably smaller.

5 5 The Fig. 5 shows results for another simulation made for the coverage area. The graph shows the behavior of for two values (0.05 Hz and 200 Hz) of the Doppler shift. The BPSK modulation was used, was set to 20 db and the SNR to 20 db. The minimum value of was 0.05 Hz, which suggests that the UAV relative position is unchanged, because in this case the velocity in negligible (near zero), and the maximum Doppler shift used was 200 Hz, equivalent to a speed of the aircraft of 108 km/h. In this simulation it can be observed that the frequency shift is proportional to the coverage area, but comparing with the results of others simulations it can be seen a much smaller influence than with the noise, for instance. Between the highest and lowest values for, the distance between the corresponding graphs is quite small. IV. CONCLUSION The general aim of this study was to estimate the coverage area and BER for a communication system for UAVs. Through simulations with the Simulink computational tool was measured what impact of noise, fading and Doppler effect they cause to the two estimates analyzed and the use of the features from ZigBee protocol allowed a comparison between its two main possibilities of modulation. Through the results it can conclude that, despite working in a lower frequency, the BPSK modulation was better than the O-QPSK for the BER and in all conditions of noise and fading. If the system requires a high frequency signal, the O- QPSK modulation should be used, but some scheme of detection and error correction should be used for the BER to remain at a satisfactory level. Finally, it realize that the Doppler effect do not causes much impact in determining the coverage area for UAV systems. This was evidenced, because even for very high speed or very low of the aircraft, the influence of the Doppler effect was very small and the values for the coverage area remained near. On the other hand, the noise and fading showed great influence on the calculation of coverage area. It was that there is a greater difficulty to estimates when these two phenomena are of great intensity. REFERENCES [1] L. A. C. Jorge, O. T. Junior, Metodologia Para Utilização de Aero Modelos em Monitoramento Aéreo, Circular Técnica, São Carlos, October [2] F. A. Medeiros, Desenvolvimento de um Veículo Aéreo não Tripulado para Aplicação em Agricultura de Precisão, Universidade Federal de Santa Maria, Santa Maria, February [3] R. B. Marques, Utilização de VANT no Auxílio à Defesa de Superfície de Base Aérea Expedionária, Instituto Tecnológico de Aeronaútica, São José dos Campos, September [4] Part 15.4: Wireless Medium Access Control (MAC)and Physical Layer (PHY) Specifications for Low Rate Wireless Personal Area Networks. (LR-WPANs), no IEEE, 2003 [5] mathworks.com, accessed on February 14, 2011 [6] M. D. Yacoub, Foundations of Mobile Radio Engineering, CRC Press, Inc., Boca Raton, Florida, [7] M. Abramowitz and I. A. Stegun, Handbook of Mathematical Functions with Formulas, Graphs and Mathematical Tables, National Bureau os Standards Applied Mathematics Series 55, Whashington, D.C., June [8] J. R. Antônio, Área de Cobertura em Ambiente de Propagação Modelado com a distribuição k-u, Instituto Nacional de Telecomunicações, Santa Rita do Sapucaí, [9] M. Petrova, J. Riihijärvi, P. Mähönen, S. Labella, Performance Study of IEEE Using Measurements and Simulations Proceedings of IEEE WCNC, Las Vegas, April [10] [11] K. M. Trevizani, Uma Extensão do Sistema de Telemetria e Telecomandos do Projeto ARARA para Transmissão Digital de Vídeo e Dados, Instituto de Ciências Matemáticas e de Computação de São Carlos, Abril ACKNOWLEDGMENT The authors are grateful to the National Institute of Science and Technology Critical Embedded Systems (Instituto Nacional de Ciência e Tecnologia Sistemas Embarcados Críticos INCT-SEC), the Araucaria Foundation and the National Council for Scientific and Technological Development (Conselho Nacional de Desenvolvimento Científico e Tecnológico CNPq) by support and encouragement.