Publikationer

With the rapid growth of the Internet of Things (IoT) applications in smart regions/cities, for example, smart healthcare, smart homes/offices, there is an increase in security threats and risks. The IoT devices solve real-world problems by providing real-time connections, data and information. Besides this, the attackers can tamper with sensors, add or remove them physically or remotely. In this study, we address the IoT security sensor tampering issue in an office environment. We collect data from real-life settings and apply machine learning to detect sensor tampering using two methods. First, a real-time view of the traffic patterns is considered to train our isolation forest-based unsupervised machine learning method for anomaly detection. Second, based on traffic patterns, labels are created, and the decision tree supervised method is used, within our novel Anomaly Detection using Machine Learning (AD-ML) system. The accuracy of the two proposed models is presented. We found 84% with silhouette metric accuracy of isolation forest. Moreover, the result based on 10 cross-validations for decision trees on the supervised machine learning model returned the highest classification accuracy of 91.62% with the lowest false positive rate.

Deep learning (DL) using large scale, high-quality IoT datasets can be computationally expensive. Utilizing such datasets to produce a problem-solving model within a reasonable time frame requires a scalable distributed training platform/system. We present a novel approach where to train one DL model on the hardware of thousands of mid-sized IoT devices across the world, rather than the use of GPU cluster available within a data center. We analyze the scalability and model convergence of the subsequently generated model, identify three bottlenecks that are: high computational operations, time consuming dataset loading I/O, and the slow exchange of model gradients. To highlight research challenges for globally distributed DL training and classification, we consider a case study from the video data processing domain. A need for a two-step deep compreEsudan method, which increases the training speed and scalability of DL training processing, is also outlined. Our initial experimental validation shows that the proposed method is able to improve the tolerance of the distributed training process to varying internet bandwidth, latency, and Quality of Service metrics.

The growing population and demand for new public buildings contribute to increased energy consumption and greenhouse emiEsudans. In Sweden, the largest amount of energy is consumed in school buildings, i.e., where schools form the highest number of public properties (30 million m ² ). In total, schools consumed 4 222 GWh of district heating and about 3 GWh of electricity for heating and other purposes in 2020. These figures lead to the realization of the need to apply effective measures to meet the European Green Deal target for 2030. Accurately forecasting energy usage is important for all stakeholders to conduct economic analysis and optimize decision-making. It is equally important in maintenance operations to allocate resources and enable the staff and students to adjust their behaviours and address the issues in buildings where peak forecasts occur. This paper develops and evaluates a power and district heating consumption for a single day and multiple days forecasting using Multivariate Recurrent Neural Network (RNN) -Long-Short term memory (LSTM) and convolutional neural networks (CNNs) and Autoencoders (AE), using daily real consumption data of six public schools provided by Skelefteå municipality in Sweden. The experimental results demonstrate that the hybrid model CNN-LSTM has achieved good accuracy compared to others, with RMSE and nRMSE error between 18%-25% and 5%-6% for electricity, respectively, and between 20%-30% RMSE and 5% nRMSE for district heating.

The development of accurate energy prediction models plays a significant role in achieving sustainability in smart cities. However, stakeholders such as municipalities face the problem of creating individual energy forecasting models for multiple building fleets which leads to an increased amount of computational resources and time spent to prepare each model. This research proposes a method using Hierarchical clustering with dynamic time warping to group similar buildings according to their consumption values and the integration of transfer Learning (TL) to share the model weights from a source building to other target buildings. Different TL models using only 20%, 40%, and 60% of the target data were tested against a standard workflow without TL for predicting electricity and district heating for several school buildings using a Multivariate LSTM model. The results show a small variation between the TL and the standard models; when trained on only 40% of the data, the models achieved an average of 0.24% RMSE improvement for district heating and a 1.23% for electricity, indicating a potential for reduced data requirements without sacrificing predictive accuracy and demonstrating TL’s efficiency to streamline the energy forecasting process for building fleets.5% nRMSE for district heating.