Photonics-based neural networks promise to outperform electronic counterparts, accelerating neural network computations while reducing power consumption and footprint. However, these solutions suffer from physical layer constraints arising from the underlying analog photonic hardware, impacting the resolution of computations (in terms of effective number of bits), requiring the use of positive-valued inputs, and imposing limitations in the fan-in and in the size of convolutional kernels. To abstract these constraints, in this paper we introduce the concept of Photonic-Aware Neural Network (PANN) architectures, i.e., deep neural network models aware of the photonic hardware constraints. Then, we devise PANN training schemes resorting to quantization strategies aimed to obtain the required neural network parameters in the fixed-point domain, compliant with the limited resolution of the underlying hardware. We finally carry out extensive simulations exploiting PANNs in image classification tasks on well-known datasets (MNIST, Fashion-MNIST, and Cifar-10) with varying bitwidths (i.e., 2, 4, and 6 bits). We consider two kernel sizes and two pooling schemes for each PANN model, exploiting $$2\times 2$$2 × 2 and $$3\times 3$$3 × 3 convolutional kernels, and max and average pooling, the latter more amenable to an optical implementation. $$3\times 3$$3 × 3 kernels perform better than $$2\times 2$$2 × 2 counterparts, while max and average pooling provide comparable results, with the latter performing better on MNIST and Cifar-10. The accuracy degradation due to the photonic hardware constraints is quite limited, especially on MNIST and Fashion-MNIST, demonstrating the feasibility of PANN approaches on computer vision tasks.

Trials and deployments of sixth Generation (6G) wireless networks, delivering extreme capacity, reliability, and efficiency, are expected as early as 2030. Attempts from both industry and academia are trying to define the next generation network infrastructure. 6G will set in motion the fourth industrial revolution, delivering global, integrated intelligence. In this scenario, Artificial Intelligence (AI)-assisted architecture for 6G networks will realize knowledge discovery, automatic network adjustment and intelligent service provisioning. The long-term vision for implementing 6G security is to implement an autonomous, self-learning AI-assisted architecture that can perform threat mitigation without disrupting the normal use, enabling the resilience and reliability of the network and fulfilling security automation. This work proposes a first implementation of a proactive threat discovery service to be deployed at 6G base stations, paving the way for collective network intelligence in the context of cybersecurity mechanisms. Specifically, a fully unsupervised Deep Learning (DL) model is presented, able to recognize both Denial of Service (DoS) Hulk and DoS Goldeneye, with 97.0% and 92.2% mean F1-score respectively.

Forthcoming 6G/NextG networks highlight the need for advanced Artificial Intelligence (AI)-based security mechanisms to identify malicious activities and adapt to emerging threats. In this context, the integration of computer vision techniques into the cybersecurity field is promising due to their potential for sophisticated pattern recognition. In this paper we introduce a computationally efficient classification scheme acting directly on the raw packets collected at base stations and enforcing real-time conversion of packets into images. The innovative points of the proposed solution are the lightweight implementation, aligning well with the demands of future 6G networks, and the operation at network edge, enabling early threat identification as close as possible to the packet origin. We investigate the performance of this approach both in terms of F1-score and prediction time using state-of-the-art computer vision architectures and a customized Convolutional Neural Network (CNN) in an intrusion detection problem using a 5G dataset. Experimental results show the superiority of the CNN architecture over complex models. Across multiple packet window sizes N (i.e., 10, 50, 100 packets), the CNN consistently outperforms the other state-of-the-art computer vision models, achieving very high F1-scores (namely, 0.99593, 0.99860, 0.99895). A scalability analysis highlights a trade-off between CNN scalability and performance, where larger N values lead to increased prediction time. On the other hand, the other computer vision models exhibit better scalability, enabling an optimal model selection without trade-offs.

In-network function offloading using programmable data plane languages like P4 offers computational resource savings and efficient operations at the network edge. However, the deployment of ML functions in P4 switches exploiting no additional hardware remains a challenge. In this paper, we tackle the challenges in deploying Deep Neural Networks (DNNs) within programmable network devices, introducing a novel distillation based on Look-Up Tables (LUTs). The proposed method maps quantized DNNs into a cascaded arrangement of LUTs without loss in accuracy and independently of the quantized network depth. The developed approach is demonstrated in two network function use cases: a cyber security use case focused on mitigating Distributed Denial of Service attacks and a malicious activity classification task in IoT Networks. Experimental results show a trade-off between model’s accuracy, expressed in terms of F1-Score and the computational demands, influenced by bit size and number of LUTs. In addition, the latency for deploying these models ranged from 54ns to 112ns, showing the method’s practical applicability in network functions.

This paper introduces a stretching strategy for nonlinear activation functions aimed to enhance noise resilience in photonic-aware neural networks. Its effectiveness is numerically demonstrated in counteracting different noise levels in low-resolution operations.