DETECTION AND LOCATION OF ANONYMOUS SIGNAL USING SENSOR NETWORK SAVITRI BEVINAKOPPA, MANIKANT BAILE, AVINASH MUTTHUN AKUMALLA Melbourne Institute of Technology 388 Lonsdale St, Melbourne, VIC 3001 AUSTRALIA sbevinakoppa@mit.edu.au, MIT121577@stud.mit.edu.au,MIT121004@stud.mit.edu.au http://www.mit.edu.au/ Abstract: - In this digital era, there are numerous users are connecting to various digital systems such as satellite, internet, networks etc. The information used in this area should be secured enough for the users. Detecting and locating an anonymous deterministic signal within an area is quite challenging. The aim of this paperis to apply Generalized Likelihood Ratio Test (GLRT) technique to detect the unknown source by using simulation tools. This paper gives analysis of simulated results of various tools: Matlab, OPNET and Celplanner. Results from these tools show that proposed signal processing technique is effective at detecting several signals. Key-Words: -Sensor network, Matlab, OPNET, Celplanner, Simulation 1 Introduction The detecting and locating anonymous signals or unauthorized objects in a Restricted Area (RA) is challenging. To secure the data, it is necessary to find anonymous signals and unidentified objects. To detect and locate an anonymous signalwithin a Restricted Area has been proposed in Generalized Likelihood Ratio Test (GLRT) technique [1]. To find anonymous signals or unauthorized objects the likelihood ratio theory is used. It is the major test that detects the parametric interruptions such as distance frequency etc. But, there is no appropriate method for non-parametric interruptions such as atmospheric conditions, terrain [2] and based on the threshold values. Even if they exist it is very difficult to optimize the values. The likelihood test theory has drawbacks. So the Generalized Likelihood Ratio Test(GLRT) has been discovered to rectify the drawbacks of nonparametric interruptions [1]. Background of the work is given in section 2.Simulation tools are used to simulate the GLRT framework using different tools such as Matlab, OPNET, Celplanner, and analysis of the results are outlined in section 3, 4 and 5. Conclusion is given in section 6. 2 Background The generalized likelihood ratio test (GLRT) is used in hypothesis testing problems [3][4] and to optimize the nonparametric values. GLRT has two distributions under null hypothesis [5]. It has various models such as nonparametric regression models, which vary with the coefficient models and generalized coefficient models, Gaussian White noise models [4]. In GLRT, simple choice of adaptive smoothing parameter is used and finite state sources are estimated and published. These Models are mostly referred for the nonparametric testing problems based on function estimation [6]. The GLRT is mostly used in the fields of real applications like satellite detection (mobile network location), radar for the military applications. The key aspiration in this paper is to track a device, an object or an anonymous signal, using binary hypothesis test and the signal is detected and the position is analyzed using a set of sensors [7]. GLRT with other features of detection technique has been simulated using Matlab, OPNET and Celplanner, to analyse the optimum technique to detect the unknown signal. The experimentations are explained in the following sections. 3 Matlab Simulation In Matlab, parameters used are: frequency, height of the radar, distance and location of signal [8][9]. The equation 1 is used for calculation of frequency. ISBN: 978-960-474-350-6 86
Simulated results for frequency F = 100MHZ, 150MHZ, 200MHZand 1000 number of sensors are shown in Fig. 1, 2 and 3. First signal shows original signal transmitted to all sensors, second signal is received signal from sensors. Fig. 1Signals with noise and without noise for F = 100MHZ Fig. 4 shows the graph of signal strength for various number (from 1000 to 80) of sensors used. This result shows that the signal strength drops as the number of sensors decreases. 4 OPNET Simulation OPNET IT Guru has been used to simulate the network traffic and time. Parameters used for OPNET simulation are:node transmission power, operational mode, data rate, dhcp, eigrp(enhanced integrated gateway routing protocol), igrp(integrated gateway routing protocol), ip(internet protocol), manet, MPLS, mobile IP, simulation efficiency, traffic and wlan. 4.1 Experiment 1 Access port is base stattion for detection of signals.mobile units are used as sensors for detection of signals.signals are deteted using this network [7]. In this experiment, about 20 sensors are used with more than 100 sources. Fig. 2Signals with noise and without noise for F = 150MHZ Fig. 5OPNET Network Fig. 3Signals with noise and without noise for F = 200MHZ Fig. 4 Result of signal strength versus sensors. Fig. 6Location of anonymous signal ISBN: 978-960-474-350-6 87
In the above network Fig. 9, mobile stations about 7 are used as sensors and the anonymous signal can be detected using mobile stations. The IP backbone [9] is used as base station to send the radio frequency signals to find the objects and the sensors gives the exact location of the object. Fig. 7 Traffic intensity Fig. 10Location of anonymous signal Fig. 8Original signal with all parameters Fig. 6,7,9 and 10 are varied according to the give parameters. When the EIGRP and IGRP are varied the graph is almost the same. Graph with increased traffic intensity shows the large variations. Efficiency also varies based on traffic intensity. The last graph shows the original signals. The duration also plays a crucial role in the network. Graphs are varied in according to hour, day and week[8]. 4.2 Experiment 2 Fig. 9 Limited number of sensors Fig. 11Original signal with all parameters In Fig. 10 and 11, simulation speed is varied in accordance with the average simulation speed. The following parameters are used for the simulation purpose. Fig 10, duration of simulation is one hour and the values per statistic are 100. a. Events: Total (154,039); Average Speed (425,477 events/sec.) b. Time: Elapsed (0.36 sec.); Simulated (1 hr. 0 min. 0 sec.) Fig. 11, simulationduration is one week and the Values per statistic are 200. c. Events: Total (26,170,847); Average Speed (1,234,000 events/sec.) d. Time: Elapsed (21 sec.); Simulated (168 hr. 0 min. 0 sec.) ISBN: 978-960-474-350-6 88
5 Celplanner Simulation Celplanner gives the coverage pattern of all sensors. Parameters used are Best server, Co-channel interference, Number of servers, Individual dbu contours, Morphology, prediction, Path profile, Link budget, Population, Project tree, Legendary, 3d view, kml setup, History, Sensors capacity (coverage), Composite reverse and Composite forward. Fig. 12 shows the sensor network used for simulation, where dense sensors are placed in the middle of the network as it is a prime location to be protected. Fig 13 and Fig. 14 shows topology and morphology respectively. Figures used for morphology are given n Fig.15. Fig. 15Morphology prediction Prediction analysis Screen shots of 3d view of sensors placed in a geographic area. Fig. 16 and 17 show population of sensors in Terrain and plain areas. Fig. 12Frame work of Sensor network Fig. 16Terrain surface with sensors Fig. 13Frame work of topology Fig. 17Plain Area Fig. 14Cell planner in morphology Fig. 18 shows that maximum coverage of the each sensor and the range of BTS are covered well. Parameters for link budget for base station and sensors are given in Fig. 19 and 20 respectively. ISBN: 978-960-474-350-6 89
Fig. 18Sensors capacity (coverage) Fig. 21Individual coverage of sensors Composite forward from base station to sources and composite reverse traffic from sources to sensor are given in Fig. 22 and 23 respectively. These 2 figure shows that there is good traffic flow from each sensor to source and vice versa. Fig. 19Link budget for base station Fig. 22Composite forward Fig. 23Composite reverse Fig. 20Link budget for sensor Coverage of each individual sensor is indicated in Fig. 21, which indicates there is a good coverage of each sensor, not overlapping with coverage of neighbour sensor. Traffic from Best server as shown in Fig. 24 shows that there is heavy traffic in the middle of the network, where many users are using network in a prime location. Fig. 25 shows indusial dbu contours. Path profile of one of the sensor ie 29, base station 1 and base station 2 are given in Fig. 26, 27 and 28 respectively. ISBN: 978-960-474-350-6 90
Fig. 28Path Profile for base station2 Fig. 24Best Server The parameters run on each of sensor as shown in Fig. 29, using a centralized sensing system the probable position of the signals sources is found. Fig. 25Indusial dbu contours Fig. 29History of the project: Fig. 26Path profile for sensor29 6. Conclusion From the simulators such as Matlab and Celplaner it is possible to detect the objects in any place using a base station and sensors. In the Celplaner more cell sites can be placed from the site and populating them. Once all the predictions are updated, we can detect the signal from the base station spreading to the cell sites and can obtain the output. From Matlab the difficulty is to predict exact threshold level for detection, if found then it is compared with energy of noise to give the perfect output signal. Out future work will focus on prediction of threshold value for detection of unknown signal. Fig. 27Path Profile for base station1 References: [1] M. S. Brandstein and D. B. Ward, Microphone Arrays: Signal Processing Techniques and Applications, Springer - Verlag, 2001, pp.322-413. ISBN: 978-960-474-350-6 91
[2] C. Marro, Y. Mahieux, and K. U. Simmer, Analysis of noise reduction and dereverberation techniques based on microphone arrays with postfiltering, IEEE Trans. Speech Audi, 2002, pp.59-98. [3] K. U. Simmer, J. Bitzer, and C. Marro, M. S. Brandstein and D. B. Ward, Post-filtering techniques, Microphone Arrays: Signal Processing Techniques and Applications, :Springer-Verlag2001, pp.39-60. [4] I. Cohen and B. Berdugo, Microphone array post-filtering for nonstationary noise suppression, Proc. ICASSP, 2002, pp.901-904. [5] S.Gannot, D. Burshtein, and E. Weinstein, Signal enhancement using beamforming and nonstationarity with applications to speech, IEEE Trans. Signal Processing, vol. 49, 2001, pp.1614-1626. [6] J. He, Z. Liu, Linearlyconstrained minimum - normalised variance beamforming against heavy-tailed impulsive noise of unknown statistics, IET Radar, Sonar & Navigation. Dec2008, Vol. 2 Issue 6, pp. 449-457. [7] Technion&mdash, Israel Institute of Technology. Yücek and HüseyinArslan, A Survey of Spectrum Sensing Algorithms for Cognitive Radio Applications, IEEE Commun. Surveys & Tutorials, Vol. 11, No.1, 2009, pp. 120-125. [8] N. K. Nahar, B. Raines, R. G. Rojas, B. Strojny, Wideband antenna array beam steering with free-space optical true-time delay engine, IET Microwaves, Antennas & Propagation. June 2011, Vol. 5 Issue 6, pp. 740-746. [9] I. Cohen and B. Berdugo, Noise estimation by minima controlled recursive averaging for robust speech enhancement, IEEE Signal Processing Lett., vol. 9, 2002, pp.12-15. [10] I. Cohen, Multi-channel post-filtering in nonstationary noise environments, Technion&mdash, Israel Institute of Technology,2002, pp.169-211. ISBN: 978-960-474-350-6 92