Emerging agricultural internet-of-things (Ag-IoT) is increasing the efficiency of farming. The data collected by the wireless-enabled Ag-IoT infrastructure is highly sensitive as corrupting the data can cause significant damages to farm production and the livelihood of growers. The trust of the data can be established by initial secure bootstrapping of the wireless underground end nodes. This paper tackles the problem of scalable and secret-free trust-establishment for commercial off-the-shelf (COTS) underground nodes with an aboveground gateway applicable to heterogeneous end nodes. Secure bootstrapping requires authentication and secret establishment, which are achieved in-band, aided by a trusted underground node by exploiting the unique and hard-to-forge underground wireless signal propagation laws. The secret-free trust-establishment for underground wireless networks (STUN) protocol is resistant to active signal injection attacks and is scalable with an increasing number of underground nodes. Further, it is theoretically proven that STUN has security equivalent to the unbalanced oil and vinegar scheme in public cryptography and validating with experiments using an underground wireless testbed.

The use of unmanned aerial vehicles (UAVs) for diverse activities has increased rapidly in recent years. Nonetheless, if operational cyber security is not handled effectively, these technologies offer a significant hazard which can cause catastrophic harm. Therefore it is important to identify the potential attacks that can be implemented by an adversary. Traditional methods for data integrity designed for the Internet are not suitable for UAV assisted vehicular or wireless sensor networks due to the high communication overhead and latency required. This paper proposes a lightweight data integrity technique called PIC to address this problem. Every device, at regular, generates an authentication parameter that depends on the packets transmitted. The authentication parameters are only delivered to a central server or the destination device where the integrity of the packets is verified. The proposed algorithm, i.e., PIC, does not include computationally expensive cryptographic algorithms. Therefore, the overhead introduced by embedding message authentication code (MAC) to every transmitted packet is significantly reduced. A formal security proof is presented in this paper to demonstrate the robustness of PIC, i.e., the PIC can identify malicious UAVs effectively. A performance analysis using NS-3 is also presented to show that PIC detects malicious UAVs with minimum communication overhead and latency.

In the scenario of highly mobile communication of Internet of Vehicles (IoV), identity authentication between vehicles not only needs to ensure security, but also needs to consider the problem of delay. The designed authentication scheme should meet the requirements of ultra-reliable and low latency in future B5G or 6G. We innovatively used the concept of network slicing to divide the interior of the vehicle into multiple compartments, and carried out PBFT consensus of blockchain among these compartments. Meanwhile, we introduce the channel estimation method to evaluate the relationship between the success rate of authentication and the time delay, which provides a reference for the deployment of PBFT consensus in IoV communication in the future.

In this paper, we propose a cognitive radio (CR)- based secure resource allocation scheme for device-to-device (D2D) communication networks. It aims to minimize the interference effect with proper pairing of strong and weak D2D users by sensing the available spectrum of cellular user (CU's). It maximizes the total sum rate of D2D user with great quality of service and quality of experience. Further, to improve the secrecy capacity of the proposed scheme, we formulate a coalition game. It has preference order (decides upon the channel conditions) and based on that it moves D2D users from one coalition to another. This landed Finally, it shift D2D users to the superlative coalition having conducive channel conditions. We then evaluate the proposed scheme considering parameters such as total sum rate, sum secrecy capacity, and average secrecy capacity. It is observed from graphs that the proposed scheme outperforms the traditional without CR, first order algorithm (FOA), and nearest first approaches.

An intelligent transportation system plays a key role in rolling out smart cities. During traveling, passengers would like to exploit the vehicular potential as a mobile resource for services including data communications, cloud storage, computing, and energy supply. By leveraging the mobile edge computing infrastructure, the connected vehicles and roadside wireless infrastructure can facilitate cooperative environmental sensing while meeting passenger needs. However, the inter-connected complex vehicle-enhanced mobile edge brings new challenges to the security and sustainability of the system.

This talk first presents our recently proposed Dynamic Bayesian Network (DBN)-based security quantification approach for the physical layer key generation pipeline, which can capitalize on the potential of the dynamic inter-vehicular channel status. Driven by real channel measurements, the proposed DBN-based solution can significantly outperform the conventional empirical ad theoretical security analysis.

The drive to reduce carbon emission coupled with over-the-roof gas prices is expediting the electrification process of vehicles; so, electric vehicles (EVs) and autonomous EVs (AEVs) are increasingly playing a significant role in meeting future transportation needs. The second part of this talk focuses on the intelligent energy management policy for the EV network with limited rechargeable batteries, roadside wireless network, and charging stations. Preliminary comparisons between the independent vehicle energy management scheme and the collaborative energy management scheme showed that the roadside system allows taxies in New York City to extend driving ranges and serve more customers.

This talk will conclude with a summary of other ongoing research projects on AEV-enhanced green mobile edge.

Cyber physical systems (CPSs) are multi-dimensional complex systems integrated with multiple computing, networking and physical elements, enabling to realize complex tasks in 5G era. Network slicing (NS) is recognized as the key enabler towards flexible allocation and sharing of resources in 5G-enabled CPSs. In NS, one crucial technical issue is the resource allocation of slice. Researchers have paid extensive efforts to study the technical issue. In real application scenarios, slice requests are generated dynamically. In addition, due to the heterogeneity nature of physical elements in 5G-enabled CPSs, security is strongly required to be considered. On these basis, secure resource allocation of dynamic slices is researched in this paper. At first, system models of CPSs (supporting NS) and slices are described. Then, the dynamic and secure resource allocation framework, abbreviated as \textit{SecDy-Slice}, is proposed and described. The goal of \textit{SecDy-Slice} framework is to achieve secure resource allocation per dynamic slice for NS-enabled CPSs. In order to validate the merits of the proposed framework, evaluation work is conducted, in the form of simulation.

In order to improve the security performance of the non-orthogonal multiple access (NOMA) cooperative transmission system, this paper studies an eavesdropping system based on NOMA cooperative transmission multi-slot combined reception (MSCR) strategy. The closed-form expression of ergodic secrecy rate (ESR) and optimal power allocation factor (OPAF) under the strategy are derived. The simulation results show that when existing an eavesdropping node with strong decoding capability, compared with the existing NOMA cooperative transmission strategy and direct link strategy, the MSCR-NOMA strategy has a maximum increase of 0.281 bps/Hz and 0.937 bps/Hz in ergodic secrecy rate.

Potential attacks may seriously threat the security and efficiency of the Internet of Things (IoT). Anomaly detecting methods are widely used to perceive such attacks, among which long short-term memory (LSTM) has prominent advantage in exploring the time correlation of IoT traffics to distinguish latent anomalies. However, comparing to the fully utilized time correlation of the whole traffics, the time correlation of different features in these traffics, especially of the closely related features, is usually ignored in current LSTM based anomaly detecting schemes. To address this issue and improve the performance of anomaly detection, we propose a feature-attended multi-flow long short-term memory (FAMF-LSTM) scheme, which classifies the features into several groups with a novel relation-based feature grouping model, splits the input data into multiple flows according to these groups, and learns the time correlation inner each flow as well as its impact on the output model. the experiments evaluate our proposed FAMF-LSTM based on three popular datasets. The experiment results show that FAMF-LSTM can improve the classification performance considerably in anomaly detection, and outperform standard Recurrent Neural Network (RNN) and LSTM by up to 10.68\% and 10.25\% respectively in detection accuracy.

We rely on the internet heavily for our business in this day and age. As a result, fully connected or wireless internet access is essential. 5G networks involve different stakeholders and bring several challenges because of different security requirements and measures. Deficiencies in security management between these stakeholders can lead to security attacks. One of the most well-known and important cyber-attacks today is distributed denial-of-service (DDoS). This paper focuses on identifying DDoS attacks that block network availability by flooding the victim with a large amount of unlawful traffic, saturating its capacity, and preventing legitimate data from passing through. The rise of AI in recent years has given a machine learning model enhanced DDoS detection. AI helps in improving cybersecurity posture and at the same time cybersecurity can protect the compromise in AI and ML systems. In this paper we present the hybrid threat detection mechanism using machine learning and human intelligence. The cyber threat detection mechanism is split between ML and Human intelligence (Split-ML). In this paper we analyze DDoS attacks based on temporal and threshold behavior for various network communication protocols. Empirical results demonstrate the effectiveness of the proposed architecture and model in detecting and predicting DDoS attacks in fully connected or wireless networks.

This paper proposes a zero-sum game theory and blockchain-based secure and decentralized dynamic pricing scheme for electric vehicle charging. It aims to secure data sharing between electric vehicles and charging station. We integrate the sixth-generation (6G) communication network to enable data transactions between electric vehicles and charging station with low latency and high reliability. We employ a zero-sum game theory approach to maximize the payoff of electric vehicles and charging station. The performance of the proposed system with 6G is evaluated by comparing it with 5G and 4G traditional networks. The performance evaluation of the proposed system has been analyzed with various parameters latency, profit for electric vehicles, profit for charging station, and optimal payoff of the system. The results show that the proposed system is highly secure and reliable than traditional systems.