Distributed CNN Inference on Resource-Constrained UAVs for Surveillance Systems: Design and Optimization

Abstract

Intrusion Detection Systems (IDSs) have played a significant role in detecting and preventing cyber-attacks within traditional computing systems. It is not surprising that the same technology is being applied to secure Internet of Things (IoT) networks from cyber threats. The limited computational resources available on IoT devices make it challenging to deploy conventional computing-based IDSs. The IDSs designed for IoT environments must also demonstrate high classification performance, utilize low-complexity models, and be of a small size. Despite significant progress in IoT-based intrusion detection, developing models that both achieve high classification performance and maintain reduced complexity remains challenging. In this study, we propose a hybrid CNN architecture composed of a lightweight CNN and bidirectional LSTM (BiLSTM) to enhance the performance of IDS on the UNSW-NB15 dataset. The proposed model is specifically designed to run onboard resourceconstrained IoT devices and meet their computation capability requirements. Despite the complexity of designing a model that fits the requirements of IoT devices and achieves higher accuracy, our proposed model outperforms the existing research efforts in the literature by achieving an accuracy of 97.28% for binary classification and 96.91% for multiclassification.

Pervasive AI for IoT Applications: A Survey on Resource-Efficient Distributed Artificial Intelligence

“…Recent works, e.g., [192], [244], proposed to avoid remote AI inferences and leverage the computation capacity of ground robots to accomplish the predictive tasks. However, only few works covered the distribution of the inference among flying drones, characterized by their faster navigation, higher power consumption, and ability to reach areas with high interferences (e.g., high-rise buildings) compared to ground devices [289], [290]. Moreover, recent efforts did not cover the path planning for different moving robots to complete their assigned missions, while performing latency-aware predictions.…”

A Survey on Collaborative DNN Inference for Edge Intelligence

“…Dhuheir et al [70] propose a strategy to allocate inference requests to resource constrained UAV groups to classify the captured airborne images, so as to obtain the minimum decision latency. Jouhari et al [18] propose a DNN distribution method in UAV to realize data classification in resource constrained equipment and avoid additional latency introduced by server based solution due to data communication from air to ground link. Disabato et al [71] introduce a method designed to allocate CNN execution on Distributed IoT applications.…”

“…Deep learning is one of the most popular AI technologies which brings the ability of automatical patterns recognition and detecting abnormal data from edge devices. Then the effective information extracted from the sensing data is fed back to the server for real-time prediction and decision-making, such as public transport planning [17], intelligent city monitoring [18] and forest fire early warning [19], so as to respond to the rapidly changing environment and improve the efficiency of scene application.…”

RL-DistPrivacy: Privacy-Aware Distributed Deep Inference for Low Latency IoT Systems

“…To cater with remote transmission, researchers have started to investigate the feasibility of distributing CNN networks among resource-limited IoT devices in order to jointly compute the inferences locally. The works in [13], [17], [18] presented per-layer CNN partitioning strategy and formulated the allocation approach to distribute different segments into resource constrained devices, as an optimization problem. In addition to per layer partitioning, the DNN can also be divided along the input dimension (e.g., rows or columns of feature maps).…”