August 12-13, 2023, Virtual Conference
Michael Mundt, Harald Baier, Bundeswehr University Munich, Germany
The risk of data theft increases significantly over the past years and as a consequence overwhelming damage is caused to institutions and private persons, respectively. As a result the widespread ISO standard 27001 was updated recently in October 2022 to integrate data exfiltration aspects through new security controls. Especially the new ISO 27001:2002 control 5.7 addresses threat intelligence as a key factor to protect against data theft. In this paper we review the ISO 27001:2022 with respect to data exfiltration and come up with recommendations on how the recently integrated ISO 27001:2022 controls can be used in an operational environment. Based on that, we introduce and demonstrate the effectiveness of a proactive and dynamic concept by integrating a simulation cycle into the Information Security Management System (ISMS) and using cyber threat intelligence to provide us with information about current threats. Key issues of our simulation approach are both a dynamic configuration and a proactive defence before the corresponding data exfiltration attacks actually occur. We provide a prototype for the simulation cycle based on a smart combination of established and widely accepted cyber defence tools. Within our evaluation we show the feasibility of our targeted and dynamically configurable simulation of data exfiltration threats and thus support to thwart the actual cyber attacks in advance. In all we contribute to prevent (or at least make it significantly more difficult) the threat of data exfiltration. Dynamic, threat aware and preventive cyber-defence is our objective and we provide an updated concept which integrates conclusively into an ISO 27001:2022 compliant ISMS.
ISO 27001:2022 Update, Data Exfiltration, Control 5.7 Threat Intelligence, Threat Simulation.
Yasir Nawaz, Shanghai Jiao Tong University, Hong Kong
In this paper, we propose a novel confidential mode of operation for block cipher, which is known as the dual key chaining mode (DKC) that is based on highly unpredictable value, that provide cryptographic protection for sensitive. The DKC is refinement of the work of existing mode of operation that is recommended by NIST. In addition to being efficient DKC modes, we discuss the security of DKC mode and existing modes of operations including the Cipher Block Chaining mode, Cipher Feedback mode, Output Feedback mode and Counter mode. The security levels of DKC mode are higher than those of the existing mode of operation when they face chosen plaintext attacks.
Dual Key Chaining mode, Cipher Block Chaining mode, Cipher Feedback mode, Output Feedback mode, Counter mode, chosen plaintext attack.
Waldson Rodrigues Cardoso, João Marco Silva and Admilson Ribamar Lima Ribeiro, Federal University of Sergipe, São Cristóvão, Brazil
This article highlights the development of an awareness tool in the form of an expert system to prevent social engineering attacks in public organizations. Social engineering attacks have significant consequences for organizations, resulting in security breaches, loss of confidential information, and reputation damage. While protective measures such as awareness training and security policies have been implemented, there is still room for improvement. The tool under development will empower users to identify and avoid psychological manipulation techniques used by attackers, thereby strengthening information security and mitigating associated risks. It addresses key concepts in information security and includes interactive modules based on learning theories, as well as artificial intelligence capabilities to identify vulnerabilities. Once developed and validated, it is expected that this tool will significantly contribute to awareness and protection against social engineering attacks in public organizations, enhancing information security and reducing risks.
Social Engineering Attacks, Information Security, Expert System, Awareness, Mitigation.
QIU Chao1, WANG Bei2, QIU Ying-jie2, CHEN Qi2 and ZHANG Zhuo-fan2, 1Zhejiang Provincial Hydrological Management Center, Hangzhou, china, 2Zhejiang University, Hangzhou, china
Frequent flood disasters in China severely threaten the safety of people's lives and properties, causing huge economic losses and casualties every year. cycles. With the development of computer technology, especially the improvement of computer computing capabilities, computer science and technology dominated by machine learning have been widely used. Flood forecasting models built using machine learning technology based on retained historical hydrological data have achieved rich results. However, the existing machine learning-based flood prediction models suffer from insufficient data feature mining. This phenomenon of insufficient feature mining leads to poor overall prediction accuracy of rainfall prediction models and is prone to overfitting. This study proposes a deep learning based on various datasets for predicting rainfall indicators. Firstly, the topography of Zhejiang Province was studied, and the geographic distribution of monitoring stations was clarified, and the original data was processed. Then, the improved Informer model was used, which uses the Prob -attention module to address the problem of long prediction length of existing data, and to mine temporal and spatial features. The model was used to predict the water flow data of a certain place in the future, and compared with traditional machine learning algorithms. Experimental The results show that the average prediction results RMSE, MAPE, and R2 of the model for historical multidimensional rainfall data are 0.39, 9.51%, and 0.91, respectively, which is significantly better than traditional time series prediction algorithms in terms of accuracy.
Informer model; time series data; self -attention; rainfall forecasting.