Pedro Sebastiao

Joseanne Viana, Hamed Farkhari, Pedro Sebastiao et al.

Despite the robust security features inherent in the 5G framework, attackers will still discover ways to disrupt 5G unmanned aerial vehicle (UAV) operations and decrease UAV control communication performance in Air-to-Ground (A2G) links. Operating under the assumption that the 5G UAV communications infrastructure will never be entirely secure, we propose Deep Attention Recognition (DAtR) as a solution to identify attacks based on a small deep network embedded in authenticated UAVs. Our proposed solution uses two observable parameters: the Signal-to-Interference-plus-Noise Ratio (SINR) and the Reference Signal Received Power (RSSI) to recognize attacks under Line-of-Sight (LoS), Non-Line-of-Sight (NLoS), and a probabilistic combination of the two conditions. In the tested scenarios, a number of attackers are located in random positions, while their power is varied in each simulation. Moreover, terrestrial users are included in the network to impose additional complexity on attack detection. To improve the systems overall performance in the attack scenarios, we propose complementing the deep network decision with two mechanisms based on data manipulation and majority voting techniques. We compare several performance parameters in our proposed Deep Network. For example, the impact of Long Short-Term-Memory (LSTM) and Attention layers in terms of their overall accuracy, the window size effect, and test the accuracy when only partial data is available in the training process. Finally, we benchmark our deep network with six widely used classifiers regarding classification accuracy. Our algorithms accuracy exceeds 4% compared with the eXtreme Gradient Boosting (XGB) classifier in LoS condition and around 3% in the short distance NLoS condition. Considering the proposed deep network, all other classifiers present lower accuracy than XGB.

Joseanne Viana, Hamed Farkhari, Luis Miguel Campos et al.

Unmanned aerial vehicle (UAV) systems are vulnerable to jamming from self-interested users who utilize radio devices for their benefits during UAV transmissions. The vulnerability occurs due to the open nature of air-to-ground (A2G) wireless communication networks, which may enable network-wide attacks. This paper presents two strategies to identify Jammers in UAV networks. The first strategy is based on time series approaches for anomaly detection where the signal available in resource blocks are decomposed statistically to find trend, seasonality, and residues, while the second is based on newly designed deep networks. The joined technique is suitable for UAVs because the statistical model does not require heavy computation processing but is limited in generalizing possible attack's identification. On the other hand, the deep network can classify attacks accurately but requires more resources. The simulation considers the location and power of the jamming attacks and the UAV position related to the base station. The statistical method technique made it feasible to identify 84.38 % of attacks when the attacker was at 30 m from the UAV. Furthermore, the Deep network's accuracy was approximately 99.99 % for jamming powers greater than two and jammer distances less than 200 meters.

Joseanne Viana, Hamed Farkhari, Pedro Sebastiao et al.

Synthetic datasets are beneficial for machine learning researchers due to the possibility of experimenting with new strategies and algorithms in the training and testing phases. These datasets can easily include more scenarios that might be costly to research with real data or can complement and, in some cases, replace real data measurements, depending on the quality of the synthetic data. They can also solve the unbalanced data problem, avoid overfitting, and can be used in training while testing can be done with real data. In this paper, we present, to the best of our knowledge, the first synthetic dataset for Unmanned Aerial Vehicle (UAV) attacks in 5G and beyond networks based on the following key observable network parameters that indicate power levels: the Received Signal Strength Indicator (RSSI) and the Signal to Interference-plus-Noise Ratio (SINR). The main objective of this data is to enable deep network development for UAV communication security. Especially, for algorithm development or the analysis of time-series data applied to UAV attack recognition. Our proposed dataset provides insights into network functionality when static or moving UAV attackers target authenticated UAVs in an urban environment. The dataset also considers the presence and absence of authenticated terrestrial users in the network, which may decrease the deep networks ability to identify attacks. Furthermore, the data provides deeper comprehension of the metrics available in the 5G physical and MAC layers for machine learning and statistics research. The dataset will available at link archive-beta.ics.uci.edu

Hamed Farkhari, Joseanne Viana, Pedro Sebastiao et al.

Only increasing accuracy without considering uncertainty may negatively impact Deep Neural Network (DNN) decision-making and decrease its reliability. This paper proposes five combined preprocessing and post-processing methods for time-series binary classification problems that simultaneously increase the accuracy and reliability of DNN outputs applied in a 5G UAV security dataset. These techniques use DNN outputs as input parameters and process them in different ways. Two methods use a well-known Machine Learning (ML) algorithm as a complement, and the other three use only confidence values that the DNN estimates. We compare seven different metrics, such as the Expected Calibration Error (ECE), Maximum Calibration Error (MCE), Mean Confidence (MC), Mean Accuracy (MA), Normalized Negative Log Likelihood (NLL), Brier Score Loss (BSL), and Reliability Score (RS) and the tradeoffs between them to evaluate the proposed hybrid algorithms. First, we show that the eXtreme Gradient Boosting (XGB) classifier might not be reliable for binary classification under the conditions this work presents. Second, we demonstrate that at least one of the potential methods can achieve better results than the classification in the DNN softmax layer. Finally, we show that the prospective methods may improve accuracy and reliability with better uncertainty calibration based on the assumption that the RS determines the difference between MC and MA metrics, and this difference should be zero to increase reliability. For example, Method 3 presents the best RS of 0.65 even when compared to the XGB classifier, which achieves RS of 7.22.

Joseanne Viana, Hamed Farkhari, Luis Miguel Campos et al.

When users exchange data with Unmanned Aerial vehicles - (UAVs) over air-to-ground (A2G) wireless communication networks, they expose the link to attacks that could increase packet loss and might disrupt connectivity. For example, in emergency deliveries, losing control information (i.e data related to the UAV control communication) might result in accidents that cause UAV destruction and damage to buildings or other elements in a city. To prevent these problems, these issues must be addressed in 5G and 6G scenarios. This research offers a deep learning (DL) approach for detecting attacks in UAVs equipped with orthogonal frequency division multiplexing (OFDM) receivers on Clustered Delay Line (CDL) channels in highly complex scenarios involving authenticated terrestrial users, as well as attackers in unknown locations. We use the two observable parameters available in 5G UAV connections: the Received Signal Strength Indicator (RSSI) and the Signal to Interference plus Noise Ratio (SINR). The prospective algorithm is generalizable regarding attack identification, which does not occur during training. Further, it can identify all the attackers in the environment with 20 terrestrial users. A deeper investigation into the timing requirements for recognizing attacks show that after training, the minimum time necessary after the attack begins is 100 ms, and the minimum attack power is 2 dBm, which is the same power that the authenticated UAV uses. Our algorithm also detects moving attackers from a distance of 500 m.

Joseanne Viana, Hamed Farkhari, Pedro Sebastiao et al.

Unmanned Aerial Vehicles (UAVs) face significant security risks from jamming attacks, which can compromise network functionality. Traditional detection methods often fall short when confronting AI-powered jamming that dynamically modifies its behavior, while contemporary machine learning approaches frequently demand substantial feature engineering and struggle with temporal patterns in attack signatures. The vulnerability extends to 5G networks employing Time Division Duplex (TDD) or Frequency Division Duplex (FDD), where service quality may deteriorate due to deliberate interference. We introduce a novel U-shaped transformer architecture that leverages Principal Component Analysis (PCA) to refine feature representations for improved wireless security. The training process is regularized by incorporating the output entropy uncertainty into the loss function, a mechanism inspired by the Soft Actor-Critic (SAC) algorithm in Reinforcement Learning (RL) to enable robust jamming detection techniques. The architecture features a modified transformer encoder specially designed to process critical wireless signal features, including Received Signal Strength Indicator (RSSI) and Signal-to- Interference-plus-Noise Ratio (SINR) measurements. We complement this with a custom positional encoding mechanism that specifically accounts for the inherent periodicity of wireless signals,enabling a more accurate representation of temporal signal patterns. In addition, we propose a batch size scheduler and implement chunking techniques to optimize convergence for time series data. These advancements contribute to up to a ten times improvement in training speed within the advanced U-shaped encoder-decoder transformer model introduced in this study. Experimental evaluations demonstrate the effectiveness of our entropy-based approach, achieving detection rates of 85.06% in NLoS scenarios.

Hamed Farkhari, Joseanne Viana, Luis Miguel Campos et al.

Increasing the cardinality of categorical variables might decrease the overall performance of machine learning (ML) algorithms. This paper presents a novel computational preprocessing method to convert categorical to numerical variables ML algorithms. It uses a supervised binary classifier to extract additional context-related features from the categorical values. Up to two numerical variables per categorical variable are created, depending on the compression achieved by the Principal Component Analysis (PCA). The method requires two hyperparameters: a threshold related to the distribution of categories in the variables and the PCA representativeness. This paper applies the proposed approach to the well-known cybersecurity NSLKDD dataset to select and convert three categorical features to numerical features. After choosing the threshold parameter, we use conditional probabilities to convert the three categorical variables into six new numerical variables. After that, we feed these numerical variables to the PCA algorithm and select the whole or partial numbers of the Principal Components (PCs). Finally, by applying binary classification with ten different classifiers, we measure the performance of the new encoder and compare it with the other 17 well-known category encoders. The new technique achieves the highest performance related to accuracy and Area Under the Curve (AUC) on high cardinality categorical variables. Also, we define the harmonic average metrics to find the best trade-off between train and test performances and prevent underfitting and overfitting. Ultimately, the number of newly created numerical variables is minimal. This data reduction improves computational processing time in Internet of things (IoT) devices in future telecommunication networks.