The identification and classification of plant diseases is challenging due to the complexity and variability of symptoms across different species, and the need for timely and accurate diagnosis to effective disease management. However, the existing methods of diagnosing plant diseases often require extensive expert knowledge and can be labor-intensive and time-consuming. This paper presents a novel method leveraging a dynamic self-attention mechanism within convolutional neural networks to enhance the classification accuracy of plant diseases. By allowing the mechanism to focus adaptively on the most informative parts of the input images, this method successfully detects the complex patterns and relationships within the disease symptoms. This method is evaluated by incorporating it in a model that specifically leverages the strengths of EfficientNet architecture combined with the Scaled Exponential Linear Unit (SeLU) activation function on grape leaves dataset containing various types of diseases. This model demonstrates superior performance in detecting a variety of plant diseases, surpassing existing baseline convolutional network methods in both speed and accuracy. This research not only advances the field of plant disease management with cutting-edge AI techniques but also offers a scalable and efficient tool for agriculture practitioners to combat plant diseases more effectively.

Abstract The Internet of Things (IoT) integrates more than billions of intelligent devices over the globe with the capability of communicating with other connected devices with little to no human intervention. IoT enables data aggregation and analysis on a large scale to improve life quality in many domains. In particular, data collected by IoT contain a tremendous amount of information for anomaly detection. The heterogeneous nature of IoT is both a challenge and an opportunity for cybersecurity. Traditional approaches in cybersecurity monitoring often require different kinds of data pre-processing and handling for various data types, which might be problematic for datasets that contain heterogeneous features. However, heterogeneous types of network devices can often capture a more diverse set of signals than a single type of device readings, which is particularly useful for anomaly detection. In this paper, we present a comprehensive study on using ensemble machine learning methods for enhancing IoT cybersecurity via anomaly detection. Rather than using one single machine learning model, ensemble learning combines the predictive power from multiple models, enhancing their predictive accuracy in heterogeneous datasets rather than using one single machine learning model. We propose a unified framework with ensemble learning that utilises Bayesian hyperparameter optimisation to adapt to a network environment that contains multiple IoT sensor readings. Experimentally, we illustrate their high predictive power when compared to traditional methods.

System logs are a valuable source of information for monitoring and maintaining the security and stability of computer systems. Techniques based on Deep Learning and Natural Language Processing have demonstrated effectiveness in detecting abnormal behaviour from these system logs. However, existing anomaly detection approaches have limitations in terms of flexibility and practicality. Techniques that rely on log templates such as DeepLog and LogBERT fail to capture semantic information and are unable to handle variability in log content. On the other hand, classification-based approaches such as LogSy, LogRobust and HitAnomaly require time-consuming data labelling for supervised training. In this paper, a novel log anomaly detection model named LogFiT is proposed. The LogFiT model doesn’t make use of a vocabulary of log templates and it doesn’t require any labeled data as the model only requires self-supervised training. The LogFiT model uses a pretrained Bidirectional Encoder Representations from Transformers (BERT)-based language model fine-tuned to recognise the linguistic patterns of the normal log data. The LogFiT model is trained using masked sentence prediction on the normal log data only. Consequently, when presented with the new log data, the model’s top-k token prediction accuracy serves as a threshold for determining whether the new log data deviates from the normal log data. Experimental results show that LogFiT’s F1 score exceeds that of baselines on the HDFS, BGL, and Thunderbird datasets. Critically, when variability is introduced in the log data during evaluation, LogFiT retains its effectiveness compared to that of baselines.

The network intrusion threats are increasingly severe with the application of computer supported coorperative work. Machine learning algorithms have been widely used in intrusion detection systems, including Multi-layer Perceptron (MLP). In this study, we proposed a two-stage model that combines the Birch clustering algorithm and MLP classifier to improve the performance of network anomaly multi-classification. In our proposed method, we first apply Birch or K-means as an unsupervised clustering algorithm to the CICIDS-2017 dataset to pre-group the data. The generated pseudo-label is then added as an additional feature to the training of the MLP-based classifier. The experimental results show that using Birch and K-Means clustering for data pre-grouping can improve intrusion detection system performance. Our method can achieve 99.73% accuracy in multi-classification using Birch clustering, which is better than similar researches using a stand-alone MLP model.

Smart city is one of the major Internet of Things (IoT) applications and has become an emerging paradigm with the recent advancements of IoT devices and sensors. But the heterogenous nature of a smart city IoT environment makes it vulnerable to many privacy and security concerns and introduces significant challenges for access control of IoT resources especially where access needs to be provided to third parties and external organizations. This paper proposes a new structural relationships-based access control (SRBAC) model that can be used to delegate resource access rights to users in a large scale IoT scenario like smart city while keeping the resource owner in full control. The proposed architecture uses smart contracts and public blockchain for managing access control for external users and a local off-block chain storage for managing access control for organization’s internal users and enforcing fine-grained access control for the resources.

This work proposes a novel priority based resource management framework for multimedia applications in distributed collaborative environment. Due to the dynamic nature of the collaborative environment, sometimes the network could be lightly loaded and sometimes highly congested. Moreover, application priority might be changed by the user dynamically. In this dynamic environment, Quality of Experience (QoE) of the end users would not be satisfactory without any resource management mechanism. Our proposed framework comprises of (i) a bandwidth estimation technique which efficiently calculates the total allowable bandwidth dynamically based on the network condition and the current application rate, and (ii) a priority based resource allocation algorithm which can dynamically control the individual flow rate according to the flow priority. Our extensive NS2 simulation work demonstrates the effectiveness of our proposed system.

This paper presents a novel mechanism for dynamic rate control of prioritised Voice Over IP (VoIP) traffic in real time. The system uses our proposed variable bit rate speech codec called Speex, which can dynamically adjust the encoding bit rate (and hence the voice quality) based on the feedback information about the network congestion, flow priority, and the instantaneous speech properties. Our extensive NS2 simulation results along with results from ITU-T standard of speech quality evaluation tool (PESQ) show that the proposed system indeed provides highest quality speech while maximising the bandwidth utilisation and reducing the network congestion.

Growing demand for streaming, voice and interactive gaming applications emphasize the importance of quality of service (QoS) provisioning in the Internet, particularly the need for maximum end-to-end delay guarantee. Current methods of QoS provisioning have either scalability concerns or cannot guarantee end-to-end delay with acceptable packet loss unless bandwidth is over-provisioned. While low jitter guarantee is required for streaming applications, maximum end-to-end delay is also required for VoIP and interactive games. We propose, analyze the stability and demonstrate the viability of three combined rate and end-to-end delay controls. The stability analysis is done on a fluid network model with greedy flows showing that all controls are globally asymptotically stable without information time lags and one of them is also globally asymptotically stable with arbitrary time lags; however it substantivally under-utilizes the network. Another control, which numerically demonstrates stability with arbitrary time lags, is implemented in edge and core routers of our WAN-in-Lab with long haul links. The prototype implementation confirms its viability and its advantage over the differentiated service architecture. The viability of the two other controls is shown by detailed NS2 packet-based simulations of an eight-node real core network.

This paper presents a novel mechanism for dynamically adapting the quality of congestion controlled voice over IP (VoIP) applications on the Internet in real time. The system uses our proposed variable bit rate speech codec called Speex, which can dynamically adjust the encoding bit rate (and hence the speech quality) based on both the feedback information about the network congestion and the instantaneous speech properties. Our extensive NS2 simulation results prove that the proposed system indeed provides highest quality speech while maximising the bandwidth utilisation and reducing the network congestion.

In recent years, IP based Virtual Private Networks (IP-VPNs) which enables enterprises to dynamically create and manage privately owned virtual networks over the public networking infrastructure have drawn significant attention both from researchers and industry IT stakeholders. One of the major research challenges in this dynamic scenario is to develop a fair flow control mechanism. This flow control problem could be partitioned into two sections, intra-VPN fairness control (internal control of aggregate flows) and inter-VPN fairness control (external control of individual sources by signals emanating from edge routers) [1]. External flow control becomes very complicated for Quality of Service (QoS) aware flows where each flow has a different minimum guaranteed rate. In this paper, firstly, we present the design of a novel QoS-aware external flow control algorithm for QoS aware flows in virtual private networks, and secondly, we present extensive NS-2 simulation results for our flow control algorithm. Simulation result shows that our flow control mechanism can efficiently maximise the bandwidth utilisation while maintaining guaranteed QoS for each individual flow.

Over the past decade, the problem of achieving fair bandwidth allocation on a link shared by multiple traffic flows has been extensively researched. However, as these flows traverse a computer network, they share many different kinds of resources, such as links, buffers, and router CPU. The ultimate goal should hence be overall fairness in the allocation of multiple resources rather than a single specific resource such as link bandwidth. In this paper, we present a novel scheduler, called prediction-based composite fair queuing (PCFQ), which jointly allocates the fair share of the link bandwidth and processing resources to all competing flows. We derive the worst-case delay bound, the work complexity, and the relative fairness bound for the PCFQ scheduler and show that it outperforms a system consisting of separate bandwidth and CPU schedulers. We further present simulation results which illustrate the improved performance characteristics achieved by PCFQ. We also demonstrate that our composite scheduler can be easily implemented on an off-the-shelf network processor such as the Intel IXP 2400. Experimental results from the IXP 2400 implementation highlight the effectiveness and high performance of this algorithm in a real-world system.

Over the past decade, the problem of fair bandwidth allocation among contending traffic flows on a link has been extensively researched. However, as these flows traverse a computer network, they share different kinds of resources (e.g., links, buffers, router CPU). The ultimate goal should hence be overall fairness in the allocation of multiple resources rather than a specific resource. Moreover, conventional resource scheduling algorithms depend strongly upon the assumption of prior knowledge of network parameters and cannot handle variations or lack of information about these parameters. In this paper, we present a novel scheduler called the composite bandwidth and CPU scheduler (CBCS), which jointly allocates the fair share of the link bandwidth as well as processing resource to all competing flows. CBCS also uses a simple and adaptive online prediction scheme for reliably estimating the processing times of the incoming data packets. Analytically, we prove that CBCS is efficient, with a per-packet work complexity of O(1). Finally, we present simulation results and experimental outcomes from a real-world implementation of CBCS on an Intel IXP 2400 network processor. Our results highlight the improved performance achieved by CBCS and demonstrate the ease with which it can be implemented on off-the-shelf hardware

The problem of achieving fairness in the allocation of the bandwidth resource on a link shared by multiple flows of traffic has been extensively researched over the last decade. However, as these flows traverse a computer network, they share many different kinds of resources such as links, processor cycles, buffers and battery power, a critical resource in mobile devices. The ultimate goal should hence be overall fairness in the allocation of multiple resources rather than a single specific resource such as link bandwidth. In our earlier work we have presented a novel scheduler called prediction-based composite fair queueing (PCFQ), which jointly allocates the fair share of the link bandwidth as well as processing resource to all competing flows. Our scheme also uses a simple and adaptive online prediction scheme for reliably estimating the execution times of the incoming data packets. We have demonstrated via simulation experiments that PCFQ can provide much improved quality of service (QoS) guarantees as compared to separate bandwidth and processor schedulers. With the rapid increase in the capacity of transmission links, the ease with which a scheduler can be implemented in real hardware systems gains paramount importance. In this paper we concentrate on the design and implementation of the PCFQ scheduler in a programmable router. We demonstrate that our scheduler can be easily implemented on an off-the-shelf network processor such as the Intel IXP 2400 board. We also validate our design by carrying out extensive experiments and demonstrate the improved performance achieved by the PCFQ scheduler. The experimental results from the IXP 2400 implementation highlight the effectiveness and high performance of this algorithm in a real world system

In today's Internet, packet processing may be accomplished in the routers within the data path. For efficient resource allocation in such routers, the packet scheduling schemes should consider multiple resources such as CPU and memory in addition to the bandwidth to improve overall performance. The dynamic nature of network load and the inherent unpredictability of processing times of data packets pose a significant challenge for CPU scheduling, which in turn also limits the application of well-known bandwidth scheduling algorithms. This paper presents a new composite scheduling algorithm called prediction-based composite fair queueing (PCFQ), which is designed to schedule both bandwidth and CPU resources adaptively, fairly, and efficiently among all the competing flows. PCFQ is the first scheduler to have the following properties: (1) a very new idea of combining both CPU and bandwidth scheduling functionalities under a single scheduling scheme that can provide better QoS guarantees to the data flows competing for both processing and bandwidth resources; and (2) a simple and adaptive online prediction scheme for the scheduling algorithm for reliably estimating the execution times of the incoming data packets. Through simulation and analysis we show that PCFQ can provide significantly better delay guarantees than those achievable through separate CPU and bandwidth schedulers.

In active and programmable networks, packet processing could be accomplished in the router within the data path. For efficient resource allocation in such networks, the packet scheduling schemes should consider multiple resources such as CPU and memory in addition to the bandwidth to improve overall performance. The inherent unpredictability of processing times of active packets poses a significant challenge in CPU scheduling. It has been identified that unlike bandwidth scheduling, prior estimation of CPU requirements of a packet is very difficult since it is platform dependent and it also depends on processing load at the time of execution and operating system scheduling etc. This paper presents an adaptive solution for estimating the processing requirements of active flows efficiently and accurately. The estimation process has been used in our composite scheduling algorithm called CBCS/sup WFQ/ to estimate processing requirement. The performances of the estimation process for our composite scheduler have been analyzed through simulation works.