Accuracy of the Estimation for the Number of Active Users

In the evaluation of the proposed estimation algorithm for the number of active users, six transmitters are assumed to be present in the network. The amplitude and phase mismatches between I and Q branches of their signal constellations, i.e. (ε, φ), are (−0.3,−15◦),(−0.3,15◦),(−0.1,−5◦),(0.1,5◦),(0.3,15◦),(0.3,−15◦), respectively. Two estimates are obtained for each transmitter in the observation time window. Furthermore, a Rayleigh fading channel with an exponential PDP of 100 ns RMS delay, which would cause inter-symbol interference to the training signals, is also considered. Taking the event that the estimated number of active users is not equal to six as an estimation error in the simulation, error probabilities of the proposed estimation algorithm under different channel conditions are evaluated, as shown in Fig. 4.4. Acceptable performance can be observed from the simulation results. In addition, the estimation accuracy degrades when inter-symbol interference is caused by the multipath channel.

0 5 10 15 20 25 30 10−3 10−2 10−1 100

SNR(dB)

P

Figure 4.3: MDP in differentiating two I/Q imbalance estimates of distinct transmit- ters. One transmitter has I/Q imbalance ε= 0.25 and φ= 5◦, the other hasε= 0.05 and φ = 15◦.

5 10 15 20 25 30 10−2 10−1 100

SNR(dB)

Error Probability

P_fa=0.01, Rayleigh fading 50ns delay P

fa=0.01, Rayleigh fading 100ns delay

P

fa=0.001, Rayleigh fading 50ns delay

P

fa=0.001, Rayleigh fading 100ns delay

Figure 4.4: Error probability in estimating the number of active users when there are six transmitters in the network. Their I/Q imbalances are (−0.3,−15◦), (−0.3,15◦), (−0.1,−5◦), (0.1,5◦), (0.3,15◦) and (0.3,−15◦), respectively.

1 2 3 4 5 6 7 8 9 1011 0 0.2 0.4 0.6 0.8 1

Estimated Number of Active Users

Frequency of Occurrence P fa = 0.01, SNR = 20 dB 1 2 3 4 5 6 7 8 9 1011 0 0.2 0.4 0.6 0.8 1

Estimated Number of Active Users

Frequency of Occurrence P fa = 0.01, SNR = 16 dB 1 2 3 4 5 6 7 8 9 1011 0 0.2 0.4 0.6 0.8 1

Estimated Number of Active Users

Frequency of Occurrence P_fa = 0.001, SNR = 20 dB 1 2 3 4 5 6 7 8 9 1011 0 0.2 0.4 0.6 0.8 1

Estimated Number of Active Users

Frequency of Occurrence

P_fa = 0.001, SNR = 16 dB

Figure 4.5: Statistical analysis for the estimation results when six active transmitters exist in the network.

for all the estimated numbers of active users is analyzed. Figure 4.5 demonstrates the frequency of occurrence when SNR = 16 dB and SNR = 20 dB in the Rayleigh fading channel of 50 ns RMS delay. In most of incorrect estimations, the estimated number of active users is just one away from the true value, caused by either false alarm or miss detection. Generally, this small deviation will not significantly affect perceiving the security level of the operating environment.

4.5

Summary

In this chapter, a novel device RF-DNA based estimation technique for the number of active users in a wireless network is proposed. As a typical device RF-DNA that

has device-specific nature, transmitter I/Q imbalance is exploited in the design. I/Q imbalance of a transmitter is first estimated from its transmitting signals, and then compared with the I/Q imbalance estimate of each previously identified active user through a hypothesis testing, where the Euclidean distance between the new and previous estimates is adopted as the test metric. If the distance is larger than a selected threshold, the two estimates are considered to be from different wireless devices. A new active user is claimed if and only if all the comparisons give a different- device decision. At the end of an estimation time window, the number of active users is obtained by counting all the identified transmitters. Simulation results have been provided to validate the proposed estimation technique.

Chapter 5

Anti-Eavesdropping OFDM

System with CSI-based Coordinate

Interleaving

In light of the security weaknesses of OFDM technology and its general acceptance in modern wireless communications, the security of OFDM communication systems needs to enhanced. This dissertation concentrates on the built-in security enhance- ment of wireless OFDM systems against eavesdropping. Two novel physical-layer eavesdropping prevention strategies for OFDM are proposed. This chapter provides insight into the proposed dynamic coordinate interleaving method, which can be em- ployed in most existing wireless OFDM systems. For situations that the modulated data symbols transmitted at OFDM subcarriers are not in a complex-number struc- ture, another dynamic subcarrier interleaving technique is investigated and presented in Chapter 6.

A novel anti-eavesdropping OFDM system through dynamic subcarrier coordi- nate interleaving is proposed in this chapter, by exploiting the reciprocal, location- dependent and time-varying nature of wireless channels. In the proposed OFDM system, the transmitter performs coordinate interleaving at partial OFDM subcarri-

ers, where the symbol coordinate of an OFDM subcarrier is interleaved in an oppor- tunistic manner depending on the instantaneous channel state information between the transmitter and intended receiver. More specifically, a subcarrier symbol asso- ciated with a CSI (i.e., channel gain or phase) larger than a predefined threshold is coordinate interleaved. Since wireless channels associated with each pair of users at separate locations exhibit independent multipath fading, the frequently updated coordinate interleaving pattern can only be shared between legitimate users based on channel reciprocity. Consequently, eavesdropping is prevented due to mismatched de-interleaving at the eavesdropper. Two coordinate interleaving schemes are inves- tigated by employing the subcarrier channel gain and phase in determining the inter- leaving pattern, respectively. In order to simultaneously evaluate the eavesdropping resilience and transmission reliability of anti-eavesdropping communication systems, a novel evaluation criterion, named probability of confidential transmission, is also proposed. Theoretical analysis and simulation results are provided to validate the effectiveness of the proposed anti-eavesdropping OFDM system. As confirmed by simulation results, the proposed system significantly outperforms the conventional OFDM system in terms of the probability of confidential transmission.

5.1

Introduction

Securing wireless communications is a critical challenge due to the inherent broad- cast nature of radio signal propagation. Adversaries can possibly intercept legitimate transmissions as long as they lie within the radio transmission coverage. Tradi- tional communication securing mechanisms largely rely on cryptographic techniques at upper layers of the protocol stacks, where the security is generally guaranteed by using either pre-distributed or public cryptography keys between communication nodes [88]. However, with limited randomness and potential secrecy leakage in such highly standardized practices, key distribution and protection face severe threats of being cracked. In recent years, physical layer security, exploiting the continual influx

of the situation- and user-dependent randomness from wireless multipath channels, is emerging as an effective means to complement conventional wireless security tech- niques [13]. The new physical layer security paradigms benefit from fundamental properties of wireless channels, such as reciprocity, as well as temporal and spatial variations.