Synthetic data generation has emerged as a promising approach to address challenges related to data scarcity, privacy constraints, and bias mitigation in machine learning applications. In networking, especially intrusion detection systems (IDS), access to balanced and unbiased datasets is critical for developing robust models. However, real-world network traffic datasets often exhibit significant class imbalance and encode discriminatory patterns across sensitive attributes, which can result in unfair decision-making. This paper proposes FairTabFormer, a transformer-based generative model tailored to produce realistic, fair, tabular synthetic network traffic data. Based on the GPT-2 decoder architecture, FairTabFormer integrates a fairness-aware training paradigm that penalizes disparity in prediction distributions across sensitive groups, specifically targeting protocol types in intrusion detection datasets. The model leverages a custom loss function that combines cross-entropy with fairness regularization, ensuring demographic parity in the generated data. The experimental evaluation uses the CIC IoT23 dataset, a comprehensive benchmark for IoT-based threat detection. The downstream performance is assessed using an XGBoost classifier using accuracy as the primary utility metric, while fairness is measured using the Demographic Parity Ratio (DPR) and the Demographic Parity Difference (DPD). Comparative results against state-of-the-art models such as CTGAN and REaLTabFormer demonstrate that FairTabFormer achieves superior fairness while maintaining high predictive utility. These findings underscore its potential to enhance fairness without compromising data quality.

Smart grids rely on communication networks to connect the physical devices and the control and computation technologies. The transmission of sensitive data over the network induces the possibility of leakage of private and sensitive information about various entities and components in the grid. To address this issue, this article proposes a privacy-preserving framework to enhance the privacy of smart grids integrated with software-defined networks. The proposed framework uses two privacy metrics (mutual information and differential privacy) and formulates a privacy-preserving distributed optimization algorithm with the objective of minimizing the network cost. We view the distributed optimization algorithm as an n-player, noncooperative game and provide distributed techniques to solve the optimization problem efficiently. We prove that our algorithm converges to the Nash equilibrium of the game while preserving the data privacy. We validate the performance of our approach using three IEEE bus systems and realistic Internet service provider network topology.

While in-network caching is an essential feature of Information Centric Networks (ICN) for improved content dissemination and reducing the bandwidth consumption at the core of the network, it is prone to many privacy threats. For example, an adversary can passively breach the privacy of a consumer by simply analyzing the different retrieval times for the same content. This paper aims to address this problem of timing analysis attacks by developing privacy-enhancing caching strategies. The proposed caching strategies use two privacy metrics, namely mutual information from information theory and differential privacy, and formulates a privacy enhancing distributed optimization problem with the objective of optimizing the network cost incurred. We efficiently solve the optimization problem by considering it as a$n$-player, non-cooperative game. We show that Nash equilibrium exists for this game and compute it using an iterative best response algorithm. We compare and validate the performance of our approach on realistic network topologies by comparing it with the existing approaches in literature and the global optimal solutions.

With the advancement in network infrastructure and the growth of IoT, more and more vehicular systems are getting connected. Architectures like Information Centric Networks (ICN) are being explored for vehicular communication to achieve robust content distribution in highly mobile, dynamic, and error prone domains. Consumer mobility is implicitly supported in ICN while producer mobility is an important challenge. In this paper, we design a mechanism to support producer mobility using the spatial locality of moving producers and reverse paths of data. Then, we model the consumer (requesting) application as a GI/M/c/N queue which is used to estimate system parameters like content delivery time distribution. We perform simulations using real world traces to evaluate the accuracy and working of our model.

Many next generation Internet architectures exist in the literature for addressing various issues like increasing traffic, mobility and efficient content dissemination. One such emerging fundamental design is Information Centric Networking (ICN). The Pending Interest Table (PIT) is one of the essential components of the ICN forwarding plane responsible for the stateful routing in ICN. Optimal size of the PIT is essential for the efficient performance of the network and the enhanced consumer experience. Therefore, the optimal sizing of the PIT under various network conditions is an important and challenging problem. To this end, this paper models the PIT of a router as a GI/M/c/N queue. The model has (i) a general arrival process to accommodate the diverse nature of traffic, (ii) a service time model which takes into account the caching at the content stores and the mobility of producers, and (iii) a sojourn time distribution which is used to characterize the content delivery time at the consumers. Using the GI/M/c/N queueing model, we formulate an optimization problem to minimize the PIT size while subjecting the interest drop probability to an upper bound. The accuracy of our analytical model is demonstrated using simulations on different Internet Service Provider (ISP) topologies across a wide range of system parameters.

Many future Internet architectures have been proposed to address issues like increasing traffic, mobility, efficient content dissemination and Named Data Networks (NDN) is emerging as one of the fundamental designs. With the ever-growing mobile data traffic, providing user-mobility has become a necessity. While consumer mobility is implicitly handled in NDN, producer mobility is still of one the main challenges. In this paper we propose a hop-count based forwarding strategy to support seamless producer mobility. The key idea of this strategy is: the router makes a decision based on the number of hops traveled by the interest whether to forward the interest using Forwarding Information Base (FIB) entries or broadcast it. The intuition behind this strategy being two-fold: spatial locality of producer and data packets follow the reverse path of interests in NDN. Using simulations, we evaluate the performance of our proposed approach and compare it with Neighbor Aware Interest Forwarding. We demonstrate that our proposed strategy achieves better throughput in terms of number of interests served while reducing the overall traffic generated.