 Research Article
 Open Access
Wireless InformationTheoretic Security in an Outdoor Topology with Obstacles: Theoretical Analysis and Experimental Measurements
 Theofilos Chrysikos^{1}Email author,
 Tasos Dagiuklas^{2} and
 Stavros Kotsopoulos^{1}
https://doi.org/10.1155/2011/628747
© Theofilos Chrysikos et al. 2011
 Received: 15 June 2010
 Accepted: 20 August 2010
 Published: 29 August 2010
Abstract
This paper presents a Wireless InformationTheoretic Security (WITS) scheme, which has been recently introduced as a robust physical layerbased security solution, especially for infrastructureless networks. An autonomic network of moving users was implemented via 802.11n nodes of an ad hoc network for an outdoor topology with obstacles. ObstructedLineofSight (OLOS) and NonLineofSight (NLOS) propagation scenarios were examined. Lowspeed user movement was considered, so that Doppler spread could be discarded. A transmitter and a legitimate receiver exchanged information in the presence of a moving eavesdropper. Average SignaltoNoise Ratio (SNR) values were acquired for both the main and the wiretap channel, and the Probability of Nonzero Secrecy Capacity was calculated based on theoretical formula. Experimental results validate theoretical findings stressing the importance of user location and mobility schemes on the robustness of Wireless InformationTheoretic Security and call for further theoretical analysis.
Keywords
 Outage Probability
 Rayleigh Fading Channel
 Path Loss Exponent
 Secrecy Rate
 User Movement
1. Introduction
Security has maintained, over the last decades, a key role in wireless communications. Recent published works have renewed the interest of researchers for physical layerbased security, formulating the Wireless InformationTheoretic Security (WITS) concept, opening the way for fruitful advances in both academia and industry. Wireless InformationTheoretic Security suggests that perfect secrecy [1] in wireless communication between a transmitter and a legitimate receiver in the presence of an eavesdropper (passive intruder) is achievable even when the average SignaltoNoise Ratio (SNR) of the main channel (established between the transmitter and the legitimate receiver) is less than the average SNR of the wiretap channel (established between the transmitter and the eavesdropper) if both channels are considered to be characterized by quasistatic Rayleigh fading. Thus, we are able to bypass the limitation of the classic Gaussian wiretap channel model [2–4], according to which the average SNR of the main channel had to be larger than that of the wiretap channel in order to establish Shannon's perfect secrecy.
Wireless InformationTheoretic Security can be implemented as an independent solution for security in wireless networks, or it can function in complementary fashion next to other implemented solutions [5–8]. Wireless InformationTheoretic Security key parameters such as the Probability of Nonzero Secrecy Capacity , the Outage Probability for a given target secrecy rate , and the Outage Secrecy Capacity were thoroughly discussed in [9, 10]. Its theoretical findings are extended to include use of LDPC channel coding scheme as a means of opportunistic channel sharing [11, 12]. However, the lack of experimental measurements and empirical results challenged the scheme's reliability and robustness in relation to reallife conditions and actual propagation environments.
In this paper, Wireless InformationTheoretic Security has been determined in autonomic networks by considering Rayleigh fading channels. Furthermore, a series of experimental measurements were conducted in order to provide a test bed for computation and evaluation of these fundamental metrics of Wireless InformationTheoretic Security, in the scenario of moving users in autonomic networks. An ad hoc network was set up, comprising of autonomic users (laptops connected via 802.11n embedded network adapters) moving in lowspeed fashion (thus discarding any possible Doppler spread phenomena). The average SNRs of both main and wiretap channel were acquired via appropriate equipment and the Probability of Nonzero Secrecy Capacity was calculated in order to evaluate WITS in an actual outdoor environment with ObstructedLineofSight (OLOS) and NonLineofSight (NLOS) schemes that comply with WITS main and wiretap channel assumptions (Rayleigh fading). The results demonstrated a significant impact of relative user location on the WITS reliability as a physical security solution.
The paper is structured as following. Section 2 presents the concept of Wireless InformationTheoretic Security and discusses its key parameters. Section 3 addresses a user movement scenario and its impact on the key parameters of Wireless InformationTheoretic Security, for a certain mobility model. Section 4 features the measurement topologies and the methodology of the experiment for the aforementioned case study of user movement. In Section 5, the results are discussed whereas Section 6 includes conclusions and, finally, in Section 7 open issues for future work are addressed.
2. Wireless InformationTheoretic Security
In [9, 10], a path loss exponent of was considered, based on an average path loss exponent value estimation in [13]. The channeldependent variation of the path loss exponent [14–16] in outdoor and indoor environments, depending on the various mechanisms contributing to the signal attenuation, in an obstacledense environment, was proven to largely compromise the Wireless InformationTheoretic Security scheme [17], due to the rapid decrease of the Probability of Nonzero Secrecy Capacity. In [18], the closedform expression for the Outage Secrecy Capacity was provided, allowing for the exact calculation of the maximum achievable secrecy rate for an upperbound value of Outage Probability. This was accomplished via a Taylor series approximation of the exponential function, which was proven to be reliable for realistic values of the Secrecy Rate.
In [19, 20], the impact of user location (in relation to colluding eavesdropper(s)) on WITS robustness was addressed. However, the user movement was not taken into consideration, especially in a propagation environment with obstacles, a notion that falls into place with fundamental theoretical assumption of quasistatic Rayleigh fading for the WITS scheme. Moreover, the lack of central infrastructure calls for more specific inquiry.
3. Moving Users in Autonomic Network
In [21], the impact of user mobility on the boundaries of secure communications was addressed, in relation to the boundaries of secure communication from a physical layer standpoint. More specifically, the impact of the approaching eavesdropper on the decrease of the Probability of Nonzero Secrecy Capacity and Outage Secrecy Capacity (maximum Secrecy Rate for a given threshold of Outage Probability and a given average SNR for the legitimate receiver) was examined.
The ad hoc nodes employ a mobility model that realistically simulates mission critical situations [22, 23]. Physical obstacles are an indispensable part of the area under study. The destination points are selected by the nodes randomly based on a uniform distribution. Each node can move to every point in the network area as long as it does not reside within the boundaries of an obstacle. When a destination point is chosen, the node moves its way around the obstacles following a recursive procedure. If there is an obstacle in the way, the node sets as its next intermediate destination the vertex of the obstacle's edge directly visible that is closest to the destination and repeats the same process all over again with starting point its initial position and destination the chosen vertex. Otherwise, the node follows this direct line to get to the desired destination.
where is the distance between the transmitter and the legitimate receiver after user movement, and is the distance between the transmitter and the eavesdropper after user movement as well.
where is the Outage Secrecy Capacity (maximum Secrecy Rate) after user movement, is the Outage Probability threshold (upperbound), and is the Secrecy Rate before user movement.
The above confirms that eavesdropper's movement towards the transmitter compromises the WITS scheme, as long as the legitimate receiver remains static, and eavesdropper movement does not alter the main channel characteristics. In order to provide measurements for this scenario, a testbed has been implemented so that realistic values of Probability of Nonzero Secrecy Capacity could be provided for an outdoor environment in the presence of obstacles. The topology and measurements acquisition are described in the following section.
4. Measurements Topology and Acquisition
An autonomic network consisting of three users was set up for the purposes of the experimental measurements. Three laptops equipped with embedded 802.11n wireless adapters created an ad hoc network: the first laptop served as transmitter, the second laptop was the legitimate receiver, and the third laptop was the passive eavesdropper.
Without loss of generality, the total EIRP of transmitting laptop was at 10 dBm. Both receivers (legitimate and eavesdropper) were equipped with the NetStumbler software that provides received power values for any given wireless network (802.11) in range [24]. In our scenario, both the transmitter and the legitimate receiver (quasistatic Rayleigh fading for main channel) were considered to be static, and the eavesdropper is allowed to move, in the presence of obstacles.
5. Results and Discussion
Average received power and SNR for OLOS1 (T3) scheme.
OLOS T3  Pr (nW)  Pr (dBm)  SNR (dB) 

A3 (main)  1,26  59  26 
B3 (main)  0,32  65  20 
C3 (main)  0,03162  75  10 
D3 (main)  0,03162  75  10 
X31 (eaves)  0,00631  82  3 
X32 (eaves)  0,5  63  22 
X33 (eaves)  0,5  63  22 
X33 (eaves)  10  65  20 
X35 (eaves)  2,51  56  29 
X36 (eaves)  1,58  58  27 
for OLOS1 (T3) scheme.
Pr legit. (nW)  Pr eaves. (nW)  SNR ratio 


1,26  0,00631  0,005007937  0,995 
1,26  0,5  0,396825397  0,716 
1,26  0,5  0,396825397  0,716 
1,26  10  7,936507937  0,112 
1,26  2,51  1,992063492  0,334 
1,26  1,58  1,253968254  0,444 
0,32  0,00631  0,01971875  0,981 
0,32  0,5  1,5625  0,390 
0,32  0,5  1,5625  0,390 
0,32  10  31,25  0,031 
0,32  2,51  7,84375  0,113 
0,32  1,58  4,9375  0,168 
0,03162  0,00631  0,199557242  0,834 
0,03162  0,5  15,81277672  0,059 
0,03162  0,5  15,81277672  0,059 
0,03162  10  316,2555345  0,003 
0,03162  2,51  79,38013915  0,012 
0,03162  1,58  49,96837445  0,020 
Average received power and SNR for OLOS2 (T4) scheme.
OLOS T4  Pr (nW)  Pr (dBm)  SNR (dB) 

A4 (main)  0,32  65  20 
B4 (main)  0,32  65  20 
C4 (main)  0,0158  78  7 
D4 (main)  0,32  65  20 
E4 (main)  1  60  25 
F4 (main)  0,05012  73  12 
X41 (eaves)  0,1  70  15 
X42 (eaves)  0,32  65  20 
X43 (eaves)  0,63  62  23 
X44 (eaves)  1  60  25 
X45 (eaves)  1,58  58  27 
X46 (eaves)  2,51  56  29 
Average SNR for both the main and the wiretap channel was calculated considering a noiseinterference level of 85 dBm (for all schemes), based on actual commercial (COTS) systems (802.11g WiFi) operating at the same frequency as the ad hoc 802.11n network within range. Environmental noise was considered 98 dBm (all schemes).
The notations Xxy (i.e., X31) refer to eavesdropper's locations, sampled from the trajectory of the eavesdropper's movement in each scheme. All possible combinations between main channel and eavesdropper average SNRs were considered and the respective Probability of Nonzero Secrecy Capacity has been determined.
As it can be seen from the results, average received power levels are in the nW scale. Average SNR for both main and wiretap channel range from a few dB above zero up to almost 30 dB. Therefore, the calculated values of Probability of Nonzero (strictly positive) Secrecy Capacity range from worstcase (a value of 0,003) where the WITS scheme is largely compromised ( ), up to 0,995, where .
As in the case of OLOS1 (T3) scheme, the average received power levels remains in the nW scale, with slightly lower values than the first case. This is due to the fact that whereas this is still an OLOS scenario, the existence of dense plantation (trees with large branches of leaves) that meddles with the signal path adds to the shadowing and the attenuation of the transmitted signal. This is evident in the legitimate receiver locations B4, C4, D4, and E4. As in the first OLOS scheme for location A3, locations A4 and F4 are considered to be behind the building surface in relation to the transmitter. However, the knifeedge diffraction effect deems this an OLOS case instead of a classic NLOS scheme.
for OLOS2 (T4) scheme.
Pr legit. (nW)  Pr eaves. (nW)  SNR ratio 


0,32  0,1  0,3125  0,762 
0,32  0,32  1  0,500 
0,32  0,63  1,96875  0,337 
0,32  1  3,125  0,242 
0,32  1,58  4,9375  0,168 
0,32  2,51  7,84375  0,113 
0,0158  0,1  6,32911392  0,136 
0,0158  0,32  20,2531646  0,047 
0,0158  0,63  39,8734177  0,024 
0,0158  1  63,2911392  0,016 
0,0158  1,58  100  0,010 
0,0158  2,51  158,860759  0,006 
1  0,1  0,1  0,909 
1  0,32  0,32  0,758 
1  0,63  0,63  0,613 
1  1  1  0,500 
1  1,58  1,58  0,388 
1  2,51  2,51  0,285 
0,05012  0,1  1,99521149  0,334 
0,05012  0,32  6,38467678  0,135 
0,05012  0,63  12,5698324  0,074 
0,05012  1  19,9521149  0,048 
0,05012  1,58  31,5243416  0,031 
0,05012  2,51  50,0798085  0,020 
Finally, the NLOS scenario is presented in Figure 4. The transmitter is fixed in location T5 whereas the legitimate receiver is situated in locations A5, B5, C5, and D5. All four locations comply with classic NLOS scenario, with D5 compensating for being behind the building with the fact that the transmitted signal penetrates the glass doors of front (leftside) and back entrance (rightside) of the building, thus reducing the attenuation that would be caused in the case of wall penetration.
Average received power and SNR for NLOS (T5) scheme.
NLOS T5  Pr (ρ W)  Pr (dBm)  SNR (dB) 

A5 (main)  15,8  78  7 
B5 (main)  10  80  5 
C5 (main)  100  70  15 
D5 (main)  3,16  85  0 
X51 (eaves)  10  80  5 
X52 (eaves)  31,62  75  10 
for NLOS (T5) scheme.
Pr legit. (ρ W)  Pr eaves. (ρ W)  SNR ratio 


15,8  10  0,632911  0,612 
15,8  31,62  2,001266  0,333 
10  10  1  0,500 
10  31,62  3,162  0,240 
100  10  0,1  0,909 
100  31,62  0,3162  0,760 
3,16  10  3,164557  0,240 
3,16  31,62  10,00633  0,091 
6. Conclusions
Three different case studies in consistence within the OLOS/NLOS scenario were examined for an autonomic network of lowspeed moving nodes (laptops connected via 802.11n ad hoc network). Additive noise and interference levels were considered to be 85 dBm for all scenarios, based on environmental noise assumption of 98 dBm and recorded interference from other operating 802.11g networks in the same frequency (2.4 GHz) within range. The NetStumbler software was used for acquisition of average received power levels.
The first OLOS scheme took into consideration knifeedge diffraction and obstruction of signal path whereas sampling eavesdropper locations along a movement trajectory. Average received power levels were in nW scale and all possible average SNR combinations provided calculated values of Probability of Nonzero (strictly positive) Secrecy Capacity ranging from worstcase, where the WITS scheme is compromised and deemed inappropriate, up to bestcase, where .
The second OLOS scheme took into consideration dense plantation shadowing that leads to further signal attenuation, still however in nW scale. Finally, the NLOS scheme offered classic NLOS cases and demonstrated a radical decrease in average received power values, in pW scale whereas calculated values of Probability of Nonzero (strictly positive) Secrecy Capacity still ranged from worstcase to bestcase. This leads us to the conclusion that a severe degeneration of the channel topology and characteristics does not necessarily compromise the WITS scheme in terms of Probability of Nonzero (strictly positive) Secrecy Capacity, as long as this degeneration applies for both the legitimate receiver and the eavesdropper. The most critical factor in WITS is the relative locations of both users in reference to the transmitter that holds a definitive impact on the robustness of the WITS scheme, confirming our theoretical assumptions and findings.
Average SNR and values for each scheme and overall.
Scheme  Av. main SNR (dB)  Av. eaves SNR (dB) 


OLOS1 (T3)  16,5  23  0,354 
OLOS2 (T4)  17,3  21,5  0,269 
NLOS (T5)  6,8  7,5  0,461 
Overall  13,53  17,33  0,361 
7. Future Work
The experimental measurements acquired in this work provide some more open issues for immediate research in the field of Wireless InformationTheoretic Security. The issue of shadowing needs to be furthermore inquired. Sitespecific measurements and channel modeling have led to an empirical method for calculation of shadowing deviation based on obstacles meddling with the signal path [25], providing a novel approach for an accurate largescale consideration of shadowing phenomena. The method was originally implemented for indoor topologies at 2.4 GHz but is valid for any topology and any frequency in question. This should be taken into consideration for the mathematical expressions of WITS key parameters.
In addition, as proven from the OLOS and NLOS topologies examined in this paper, interference from other operating networks in the same frequency needs to be taken into consideration in the SNR denominator. In the case of nonuniform interference for all concerned users of the network, a noiseinterference factor needs to be implemented into the mathematical expressions of WITS key parameters, and the impact of its numerical variation (for realistic scenarios) on the WITS reliability needs to be thoroughly examined.
Finally, the issue of Doppler spread should be addressed for higher values of the user velocity, where both the channel characteristics and the Secrecy Rate are affected by Doppler shift.
Declarations
Acknowledgments
The authors would like to acknowledge Mr. Giannis Georgopoulos for his assistance during the experimental work. The authors wish to acknowledge the support of the ICT European Research Programme and all the partners in PEACE: PDMF&C, Instituto de Telecomunicaes, FhG Fokus, University of Patras, Thales, Telefonica, and CeBit.
Authors’ Affiliations
References
 Shannon CE: Communication theory of secrecy systems. Bell Systems Technical Journal 1949, 29: 656715.MathSciNetView ArticleGoogle Scholar
 Wyner AD: The wiretap channel. Bell Systems Technical Journal 1975, 54(8):13551387.MATHMathSciNetView ArticleGoogle Scholar
 Csiszar I, Korner J: Broadcast channels with confidential messages. IEEE Transactions on Information Theory 1978, 24(3):339348. 10.1109/TIT.1978.1055892MATHMathSciNetView ArticleGoogle Scholar
 LeungYanCheong SK, Hellman ME: The Gaussian wiretap channel. IEEE Transactions on Information Theory 1978, 24(4):451456. 10.1109/TIT.1978.1055917MATHMathSciNetView ArticleGoogle Scholar
 Maurer UM: Secret key agreement by public discussion from common information. IEEE Transactions on Information Theory 1993, 39(3):733742.MATHView ArticleGoogle Scholar
 Maurer UM: Informationtheoretically secure secretkey agreement by NOT authenticated public discussion. In Advances in Cryptology—EUROCRYPT '97, Lecture Notes in Computer Science. Volume 1233. Springer, Heidelberg, Germany; 1997:209225.Google Scholar
 Maurer UM: Informationtheoretic key agreement: from weak to strong secrecy for free. In Advances in Cryptology—EUROCRYPT 2000, Lecture Notes in Computer Science. Volume 1807. Springer, Heidelberg, Germany; 2000:351368.Google Scholar
 Maurer U, Wolf S: Secretkey agreement over unauthenticated public channels—part I: definitions and a completeness result. IEEE Transactions on Information Theory 2003, 49(4):822831. 10.1109/TIT.2003.809563MATHMathSciNetView ArticleGoogle Scholar
 Barros J, Rodrigues MRD: Secrecy capacity of wireless channels. In Proceedings of IEEE International Symposium on Information Theory (ISIT '06), July 2006. IEEE Press; 356360.Google Scholar
 Bloch M, Barros J, Rodrigues MRD, McLaughlin SW: Wireless informationtheoretic security. IEEE Transactions on Information Theory 2008, 54(6):25152534.MATHMathSciNetView ArticleGoogle Scholar
 Bloch M, Thangaraj A, McLaughlin SW, Merolla JM: LDPCbased Gaussian key reconciliation. In Proceedings of IEEE Information Theory Workshop (ITW '06), March 2006. IEEE Press; 116120.Google Scholar
 Richardson TJ, Shokrollahi MA, Urbanke RL: Design of capacityapproaching irregular lowdensity paritycheck codes. IEEE Transactions on Information Theory 2001, 47(2):619637. 10.1109/18.910578MATHMathSciNetView ArticleGoogle Scholar
 Rappaport T: Wireless Communications: Principles and Practice. Prentice Hall, Upper Saddle River, NJ, USA; 2001.Google Scholar
 Parsons JD: The Mobile Radio Propagation Channel. Wiley Interscience, Hoboken, NJ, USA; 2000.View ArticleGoogle Scholar
 Özgür A, Lévêque O, Preissmann E: Scaling laws for one and twodimensional random wireless networks in the lowattenuation regime. IEEE Transactions on Information Theory 2007, 53(10):35733585.View ArticleGoogle Scholar
 Seybold J: Introduction to RF Propagation. Wiley Interscience, Hoboken, NJ, USA; 2005.View ArticleGoogle Scholar
 Chrysikos T, Kotsopoulos S: Impact of channeldependent variation of path loss exponent on Wireless InformationTheoretic Security. In Wireless Telecommunications Symposium 2009. IEEE Press, Prague, Czech Republic; 2009:17.View ArticleGoogle Scholar
 Chrysikos T, Dagiuklas T, Kotsopoulos S: A closedform expression for outage secrecy capacity in Wireless InformationTheoretic Security. In Proceedings of Security in Emerging Wireless Communication and Networking Systems (SEWCN '09), 2010, Lecture Notes in Computer Science. Volume 42. Springer; 312.View ArticleGoogle Scholar
 Pinto PC, Barros J, Win MZ: Physicallayer security in stochastic wireless networks. In Proceedings of 11th IEEE Singapore International Conference on Communication Systems (ICCS '08), November 2008. IEEE Press; 974979.Google Scholar
 Pinto PC, Barros J, Win MZ: Wireless physicallayer security: the case of colluding eavesdroppers. In Proceedings of IEEE International Symposium on Information Theory (ISIT '09), July 2009. IEEE Press; 24422446.Google Scholar
 Chrysikos T, Dagiuklas T, Kotsopoulos S: Wireless informationtheoretic security for moving users in autonomic networks. IFIP Wireless Days (WD '10), 2010, Venice, ItalyGoogle Scholar
 Papageorgiou C, Birkos K, Dagiuklas T, Kotsopoulos S: An obstacleaware human mobility model for ad hoc networks. Proceedings of the 17th IEEE International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '09), September 2009, London, UKGoogle Scholar
 Papageorgiou C, Birkos K, Dagiuklas T, Kotsopoulos S: Simulating mission critical mobile ad hoc networks. Proceedings of the 4th ACM International Workshop on Performance Monitoring, Measurement, and Evaluation of Heterogeneous Wireless and Wired Networks (PM2HW2N '09), October 2009, Tenerife, SpainGoogle Scholar
 http://www.netstumbler.com/
 Chrysikos T, Georgopoulos G, Kotsopoulos S: Empirical calculation of shadowing deviation for complex indoor propagation topologies at 2.4 GHz. In Proceedings of International Conference on Ultra Modern Telecommunications (ICUMT '09), October 2009, St. Petersburg, Russia. IEEE Press; 16.Google Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.