Machine Learning Cell Throughput Prediction to Enhance Handover Processes: A Hyperparameter-Tuned Approach within the ORAN Framework

Recently, there have been many attempts to apply Machine Learning (ML)-based prediction mechanisms In wireless networks. One open question is how reliable such predictions can be, and how well ML models can learn from the radio environment. In this paper, we present initial results on Quality of Service (QoS) prediction using the example of throughput prediction. We focus on suggesting new sets of features that can improve the prediction performance for different prediction horizons. Thereby, we identify important features that have a large impact when using radio environment data as input for ML models. To this end, we consider information from space, time, and network domains. In particular, we show that features, such as cell throughput and previous users’ data can significantly improve the ML model performance. Besides the importance of input features, we also investigate how the prediction performance deteriorates for different prediction horizons.

Wireless communication contains many fluctuations than wired networks. In this paper, we present several machine learning and deep learning models to predict future network throughput, which is crucial for reducing latency in online streaming services. This paper explains the main components of the throughput prediction system. The throughput prediction model includes data input, data training, and prediction computation parts. This model accepts network throughput for the training data of the model and forecasts future data. We also present the advantages and limitations of utilizing AI models for throughput prediction. Finally, we believe that this study highlights the impact of deep learning techniques for throughput prediction.

Modern-day healthcare is being challenged by the critical nature of traffic requirements for services like health monitoring, remote consultations, and robotic surgery. While the existing literature addresses various algorithms for real-world throughput prediction, there is a lack of systematic exploration of these algorithms in healthcare. In this paper, we study the problem of accurately predicting performance parameters in a smart healthcare environment. Taking into account the characteristics of healthcare networks, we compare machine learning based algorithms and prediction models for performance and performance characteristics. We designed a model based on the Healthcare 4.0 paradigm for throughput prediction of two different types of healthcare data - sensor data used for telemonitoring e.g., Blood Pressure (BP) and Sp02 sensor data, and audio/video data that is used for teleconsultation. Among the algorithms, K-means performs the best achieving 72.68% accuracy and 75.9% sensitivity, while Naive Bayes gives better precision at 72% in predicting the throughput.

Vehicle-to-Everything Communication (V2X) technologies aim to meet strict quality-of-service (QoS) requirements of vehicular connectivity applications such as safety message exchange, remote driving, and sensor data sharing. The high reliability requirement is particularly important to enable safety relevant applications. Thus, predicting QoS levels becomes key to ensure the reliability of the connected vehicle applications. Recently, machine learning (ML) algorithms are demonstrated to provide dependable predictions to plan, simulate, and evaluate the performance of vehicular networks. In this paper, we propose ML aided New Radio (NR)-V2X QoS predictions scheme to provide Packet Delivery Ratio (PDR) and throughput predictions with the input of Modulation and Coding Schemes (MCS), distance-to-base station, Signal to Interference plus Noise Ratio (SINR), and packet size. Seven different ML algorithms based prediction models are trained and evaluated by using NR-V2X simulation data. We provide performance comparisons between Support Vector Regression (SVR), Deep Neural Network (DNN), Random Forest (RF), Gradient Boosting Machine (GBM), Extreme Gradient Boosting (XGBoost), Categorical Boosting (CatBoost), and Light GBM (LGBM) for predicting throughput and PDR. We demonstrate that CatBoost and RF are the best performing algorithms to predict throughput and PDR of NR-V2X networks, respectively.

Scientific workflows are increasingly transferring large amounts of data between high performance computing (HPC) systems. Even though these HPC systems are connected via high-speed dedicated networks and use dedicated data transfer nodes (DTNs), it is still difficult to predict the data transfer throughput because of variations in data transfer protocols, host configurations, performance of file systems, and overlapping workloads. In order to provide reliable performance prediction for better resource management and job scheduling, we need models for predicting data transfer throughput under real-world conditions. In this paper, we explore different machine learning approaches for building data-driven models to improve performance and prediction of large-scale data transfer throughput. In addition to the variables already collected by the network monitoring system, we also develop heuristics to derive additional metrics for improving the prediction accuracy. We use the prediction results to identify the importance of different network parameters in predicting the throughput for large-scale data transfers. Through extensive tests, we identify key network parameters, discover interesting variations among different HPC sites, and show that we can predict throughput with high accuracy. We also analyze our models and results to provide recommendations for improving the performance of big data transfers.

In a semiconductor fab, wafer lots are processed in complex sequences with re-entrants and parallel machines. It is necessary to ensure smooth wafer lot flows by detecting potential disturbances in a real-time fashion to satisfy the wafer lots’ demands. This study aims to identify production factors that significantly affect the system’s throughput level and find the best prediction model. The contributions of this study are as follows: (1) this is the first study that applies machine learning techniques to identify important real-time factors that influence throughput in a semiconductor fab; (2) this study develops a test bed in the Anylogic software environment, based on the Intel minifab layout; and (3) this study proposes a data collection scheme for the production control mechanism. As a result, four models (adaptive boosting, gradient boosting, random forest, decision tree) with the best accuracies are selected, and a scheme to reduce the input data types considered in the models is also proposed. After the reduction, the accuracy of each selected model was more than 97.82%. It was found that data related to the machines’ total idle times, processing steps, and machine E have notable influences on the throughput prediction.

Amalgamation of mmWave and LTE sub-6 provides an advantage in enhancing the reliability, communication bandwidth and optimized coverage of intelligent network and its significant applications; however, identifying the right mmWave RRU (remote Radio units) is one of the main obstacles due to blindness coverage of directional beams. Moreover, mmWave network depends on edge cloud deployment to satisfy the low latency of smart applications. Further, the minimization of energy consumption for handover is necessary due to the limited battery for IoT-device. Hence, it is necessary to minimize the signal overhead in handover process; at first, this research work focuses on the efficient handover mechanism among the mmWave and LTE, later automated handover execution process between mmWave and LTE is developed using XGBoost classification mechanism. The XGBoost algorithm is used predict the handover success rate through channel information based on the sampling window. At last, XGBoost based handover mechanism along with the standard handover mechanism is integrated to minimize the signal overhead and improvise handover success rate. Moreover, performance evaluation is carried out through varying the IoT device and obtained result shows that XGBoost-based handover proves to be better than the existing mechanism of KNN-based handover execution algorithm.

The rapid evolution of cellular system design towards 5G and beyond gives rise to a need for investigation of the new features, design proposals and solutions in realistic settings for various deployments and use case scenarios. While many system level simulators for 4G and 5G exist today, there is particularly a dire need for a 3GPP compliant system level holistic and realistic simulator that can support evaluation of the plethora of AI based network automation solutions being proposed in literature. In this paper we present such a simulator developed at AI4networks Lab, called SyntheticNET. To the best of authors' knowledge, SyntheticNET is the very first python-based simulator that fully conforms to 3GPP 5G standard release 15 and is upgradable to future releases. The key distinguishing features of SyntheticNET compared to existing simulators include: 1) a modular structure to facilitate cross validation and upgrading to future releases; 2) flexible propagation modeling using measurement based, ray tracing based or AI based propagation modeling; 3) ability to import data sheet based on measurement based realistic vendor specific base station features such as antenna and energy consumption pattern; 4) support for 5G standard based adaptive numerology; 5) realistic and user-specific mobility patterns that are yielded from actual geographical maps; 6) detailed handover (HO) process implementation; and 7) incorporation of database aided edge computing. Another key feature of the SyntheticNET is the ease with which it can be used to test AI based network automation solutions. Being the first python based 5G simulator, this ease, in part stems for SyntheticNET's built-in capability to process and analyze large data sets and integrated access to Machine Learning libraries. Thus, SyntheticNET simulator offers a powerful platform for academia and industry alike to investigate not only new solutions for optimally designing, deploying and operating existing and emerging cellular networks but also for enabling AI empowered deep automation in the future.

5G Wireless Technology Throughput Prediction Using Ensemble Machine Learning Approach

An important aspect of managing multi access point (AP) IEEE 802.11 networks is the support for mobility management by controlling the handover process. Most handover algorithms, residing on the client station (STA), are reactive and take a long time to converge, and thus severely impact Quality of Service (QoS) and Quality of Experience (QoE). Centralized approaches to mobility and handover management are mostly proprietary, reactive and require changes to the client STA. In this paper, we first created an Software-Defined Networking (SDN) modular handover management framework called HuMOR, which can create, validate and evaluate handover algorithms that preserve QoS. Relying on the capabilities of HuMOR, we introduce ABRAHAM, a machine learning backed, proactive, handover algorithm that uses multiple metrics to predict the future state of the network and optimize the load to ensure the preservation of QoS. We compare ABRAHAM to a number of alternative handover algorithms in a comprehensive QoS study, and demonstrate that it outperforms them with an average throughput improvement of up to 139%, while statistical analysis shows that there is significant statistical difference between ABRAHAM and the rest of the algorithms.