Biblio

Emails are widely used as a form of communication and sharing files in an organization. However, email is widely used by cybercriminals to spread malware and carrying out cyber-attacks. We implemented an open-source email gateway in conjunction with a security sandbox for securing emails against malicious attachments. The email gateway scans all incoming and outgoing emails and stops emails containing suspicious files. An automated python script would then send the suspected email to the sandboxing element through sandbox API for further analysis, while the script is used also for the prevention of duplicate results. Moreover, the mail server administrator receives notifications from the email gateway about suspicious attachments. If detected attachment is a true positive based on the sandbox analysis result, email is deleted, otherwise, the email is delivered to the recipient. The paper describes in an empirical way the steps followed during the implementation, results, and conclusions of our research.

Obfuscation refers to changing the structure of code in a way that original semantics can be hidden. These techniques are often used by application developers for code hardening but it has been found that obfuscation techniques are widely used by malware developers in order to hide the work flow and semantics of malicious code. Class Encryption, Code Re-Ordering, Junk Code insertion and Control Flow modifications are Code Obfuscation techniques. In these techniques, code of the application is changed. These techniques change the signature of the application and also affect the systems that use sequence of instructions in order to detect maliciousness of an application. In this paper an ’Opcode sequence’ based detection system is designed and tested against obfuscated samples. It has been found that the system works efficiently for the detection of non obfuscated samples but the performance is effected significantly against obfuscated samples. The study tests different code obfuscation schemes and reports the effect of each on sequential opcode based analytic system.

Distributed training is a novel approach to accelerating training of Deep Neural Networks (DNN), but common training libraries fall short of addressing the distributed nature of heterogeneous processors or interruption by other workloads on the shared processing nodes. This paper describes distributed training of DNN on computational storage devices (CSD), which are NAND flash-based, high-capacity data storage with internal processing engines. A CSD-based distributed architecture incorporates the advantages of federated learning in terms of performance scalability, resiliency, and data privacy by eliminating the unnecessary data movement between the storage device and the host processor. The paper also describes Stannis, a DNN training framework that improves on the shortcomings of existing distributed training frameworks by dynamically tuning the training hyperparameters in heterogeneous systems to maintain the maximum overall processing speed in term of processed images per second and energy efficiency. Experimental results on image classification training benchmarks show up to 3.1x improvement in performance and 2.45x reduction in energy consumption when using Stannis plus CSD compare to the generic systems.

Software developers make mistakes that can lead to failures of a software product. One approach to detect defects is static analysis: examine code without execution. Currently, various source code static analysis tools are widely used to detect defects. However, source code analysis is not enough. The reason for this is the use of third-party binary libraries, the unprovability of the correctness of all compiler optimizations. This paper introduces BinSide : binary static analysis framework for defects detection. It does interprocedural, context-sensitive and flow-sensitive analysis. The framework uses platform independent intermediate representation and provide opportunity to analyze various architectures binaries. The framework includes value analysis, reaching definition, taint analysis, freed memory analysis, constant folding, and constant propagation engines. It provides API (application programming interface) and can be used to develop new analyzers. Additionally, we used the API to develop checkers for classic buffer overflow, format string, command injection, double free and use after free defects detection.

One of the major issues which are hindering widespread and seamless adoption of Internet of Thing (IoT) is security. The IoT devices are vulnerable and susceptible to attacks which became evident from a series of recent large-scale distributed denial-of-service (DDoS) attacks, leading to substantial business and financial losses. Furthermore, in order to find vulnerabilities in IoT, there is a lack of comprehensive security analysis framework. In this paper, we present a modular, adaptable and tunable framework, called PIT, to probe IoT systems at different layers of design and implementation. PIT consists of several security analysis engines, viz., penetration testing, fuzzing, static analysis, and dynamic analysis and an exploitation engine to discover multiple IoT vulnerabilities, respectively. We also develop a novel grey-box fuzzer, called Applica, as a part of the fuzzing engine to overcome the limitations of the present day fuzzers. The proposed framework has been evaluated on a real-world IoT testbed comprising of the state-of-the-art devices. We discovered several network and system-level vulnerabilities such as Buffer Overflow, Denial-of-Service, SQL Injection, etc., and successfully exploited them to demonstrate the presence of security loopholes in the IoT devices.

The Open Data Cube (ODC), created and facilitated by the Committee on Earth Observation Satellites (CEOS), is an open source software architecture that continues to gain global popularity through the integration of analysis-ready data (ARD) on cloud computing frameworks. In 2021, CEOS released a new ODC sandbox that provides global users with a free and open programming interface connected to Google Earth Engine datasets. The open source toolset allows users to run application algorithms using a Google Colab Python notebook environment. This tool demonstrates rapid creation of science products anywhere in the world without the need to download and process the satellite data. Basic operation of the tool will support many users but can also be scaled in size and scope to support enhanced user needs. The creation of the ODC sandbox was prompted by the migration of many CEOS ARD satellite datasets to the cloud. The combination of these datasets in an interoperable data cube framework will inspire the creation of many new application products and advance open science.

Machine learning inference engine is of great interest to smart edge computing. Compute-in-memory (CIM) architecture has shown significant improvements in throughput and energy efficiency for hardware acceleration. Emerging nonvolatile memory (eNVM) technologies offer great potentials for instant on and off by dynamic power gating. Inference engine is typically pre-trained by the cloud and then being deployed to the field. There is a new security concern on cloning of the weights stored on eNVM-based CIM chip. In this paper, we propose a countermeasure to the weight cloning attack by exploiting the process variations of the periphery circuitry. In particular, we use weight fine-tuning to compensate the analog-to-digital converter (ADC) offset for a specific chip instance while inducing significant accuracy drop for cloned chip instances. We evaluate our proposed scheme on a CIFAR-10 classification task using a VGG- 8 network. Our results show that with precisely chosen transistor size on the employed SAR-ADC, we could maintain 88% 90% accuracy for the fine-tuned chip while the same set of weights cloned on other chips will only have 20 40% accuracy on average. The weight fine-tune could be completed within one epoch of 250 iterations. On average only 0.02%, 0.025%, 0.142% of cells are updated for 2-bit, 4-bit, 8-bit weight precisions in each iteration.

The ability to specify various policies with different overriding criteria allows for complex sets of sharing policies. This is particularly useful in situations in which data privacy depends on various properties of the data, and complex policies are needed to express the conditions under which data is protected. However, if overriding policy decisions constrain the affected data, decisions from overridden policies should not be suppressed completely, because they can still apply to subsets of the affected data. This article describes how a privacy policy framework can be extended with a mechanism to partially override decisions based on specified constraints. Our solution automatically generates complementary sets of decisions for both the overridden and the complementary, non-overridden subsets of the data, and thus, provides a means to specify a complex policies tailored to specific properties of the protected data.

Honeypots are the information system resources used for capturing and analysis of cyber attacks. Highinteraction Honeypots are capable of capturing attacks in their totality and hence are an ideal choice for capturing multi-stage cyber attacks. The term multi-stage attack is an abstraction that refers to a class of cyber attacks consisting of multiple attack stages. These attack stages are executed either by malicious codes, scripts or sometimes even inbuilt system tools. In the work presented in this paper we have proposed a framework for capturing, analysis and modelling of multi-stage cyber attacks. The objective of our work is to devise an effective mechanism for the classification of multi-stage cyber attacks. The proposed framework comprise of a network of high interaction honeypots augmented with an attack analysis engine. The analysis engine performs rule based labeling of captured honeypot data. The labeling engine labels the attack data as generic events. These events are further fused to generate attack graphs. The hence generated attack graphs are used to characterize and later classify the multi-stage cyber attacks.

Since 2018, Thai People Map and Analytics Platform (TPMAP) has been developed with the aims of supporting government officials and policy makers with integrated household and community data to analyze strategic plans, implement policies and decisions to alleviate poverty. However, to acquire complex information from the platform, non-technical users with no database background have to ask a programmer or a data scientist to query data for them. Such a process is time-consuming and might result in inaccurate information retrieved due to miscommunication between non-technical and technical users. In this paper, we have developed a Thai conversational agent on top of TPMAP to support self-service data analytics on complex queries. Users can simply use natural language to fetch information from our chatbot and the query results are presented to users in easy-to-use formats such as statistics and charts. The proposed conversational agent retrieves and transforms natural language queries into query representations with relevant entities, query intentions, and output formats of the query. We employ Rasa, an open-source conversational AI engine, for agent development. The results show that our system yields Fl-score of 0.9747 for intent classification and 0.7163 for entity extraction. The obtained intents and entities are then used for query target information from a graph database. Finally, our system achieves end-to-end performance with accuracies ranging from 57.5%-80.0%, depending on query message complexity. The generated answers are then returned to users through a messaging channel.

In this research we propose a framework that will strengthen the IoT devices security from dual perspectives; avoid devices to become attack target as well as a source of an attack. Unlike traditional devices, IoT devices are equipped with insufficient host-based defense system and a continuous internet connection. All time internet enabled devices with insufficient security allures the attackers to use such devices and carry out their attacks on rest of internet. When plethora of vulnerable devices become source of an attack, intensity of such attacks increases exponentially. Mirai was one of the first well-known attack that exploited large number of vulnerable IoT devices, that bring down a large part of Internet. To strengthen the IoT devices from dual security perspective, we propose a two step framework. Firstly, confine the communication boundary of IoT devices; IoT-Sphere. A sphere of IPs that are allowed to communicate with a device. Any communication that violates the sphere will be blocked at the gateway level. Secondly, only allowed communication will be evaluated for potential attacks and anomalies using advance detection engines. To show the effectiveness of our proposed framework, we perform couple of attacks on IoT devices; camera and google home and show the feasibility of IoT-Sphere.

In the current scenario, the data owners would like to access data from anywhere and anytime. Hence, they will store their data in public or private cloud along with encryption and particular set of attributes to access control on the cloud data. While uploading the data into public or private cloud they will assign some attribute set to their data. If any authorized cloud user wants to download their data they should enter that particular attribute set to perform further actions on the data owner's data. A cloud user wants to register their details under cloud organization to access the data owner's data. Users wants to submit their details as attributes along with their designation. Based on the Users details Semi-Trusted Authority generates decryption keys to get control on owner's data. A user can perform a lot of operation over the cloud data. If the user wants to read the cloud data he needs to be entering some read related, and if he wants to write the data he needs to be entering write related attribute. For each and every action user in an organization would be verified with their unique attribute set. These attributes will be stored by the admins to the authorized users in cloud organization. These attributes will be stored in the policy files in a cloud. Along with this attribute,a rule based engine is used, to provide the access control to user. If any user leaks their decryption key to the any malicious user data owners wants to trace by sending audit request to auditor and auditor will process the data owners request and concludes that who is the convict.

There is very significant role of Virtualization in cloud computing. The physical hardware in the cloud computing reside with the host machine and the virtualization software runs on it. The virtualization allows virtual machines to exist. The host machine shares its physical components such as memory, storage, and processor ultimately to handle the needs of the virtual machines. If an attacker effectively compromises one VM, it could outbreak others on the same host on the network over long periods of time. This is an gradually more popular method for cross-virtual-machine attacks, since traffic between VMs cannot be examined by standard IDS/IPS software programs. As we know that the cloud environment is distributed in nature and hence more susceptible to various types of intrusion attacks which include installing malicious software and generating backdoors. In a cloud environment, where organizations have hosted important and critical data, the security of underlying technologies becomes critical. To alleviate the hazard to cloud environments, Intrusion Detection Systems (IDS) are a cover of defense. In this paper, we are proposing an innovative model for Intrusion Detection System for securing Host machines in cloud infrastructure. This proposed IDS has two important features: (1) signature based and (2) prompt alert system.

Referring expression comprehension (REF) aims at identifying a particular object in a scene by a natural language expression. It requires joint reasoning over the textual and visual domains to solve the problem. Some popular referring expression datasets, however, fail to provide an ideal test bed for evaluating the reasoning ability of the models, mainly because 1) their expressions typically describe only some simple distinctive properties of the object and 2) their images contain limited distracting information. To bridge the gap, we propose a new dataset for visual reasoning in context of referring expression comprehension with two main features. First, we design a novel expression engine rendering various reasoning logics that can be flexibly combined with rich visual properties to generate expressions with varying compositionality. Second, to better exploit the full reasoning chain embodied in an expression, we propose a new test setting by adding additional distracting images containing objects sharing similar properties with the referent, thus minimising the success rate of reasoning-free cross-domain alignment. We evaluate several state-of-the-art REF models, but find none of them can achieve promising performance. A proposed modular hard mining strategy performs the best but still leaves substantial room for improvement.

Self-sovereign identities provide user autonomy and immutability to individual identities and full control to their identity owners. The immutability and control are possible by implementing identities in a decentralized manner on blockchains that are specially designed for identity operations such as Hyperledger Indy. As with any type of identity, self-sovereign identities too deal with Personally Identifiable Information (PII) of the identity holders and comes with the usual risks of privacy and security. This study examined certain scenarios of personal data disclosure via credential exchanges between such identities and risks of man-in-the-middle attacks in the blockchain based identity system Hyperledger Indy. On the basis of the findings, the paper proposes the following enhancements: 1) A novel attribute sensitivity score model for self-sovereign identity agents to ascertain the sensitivity of attributes shared in credential exchanges 2) A method of mitigating man-in-the-middle attacks between peer self-sovereign identities and 3) A novel quantitative model for determining a credential issuer's reputation based on the number of issued credentials in a window period, which is then utilized to calculate an overall confidence level score for the issuer.

Technology advancement also increases the risk of a computer's security. As we can have various mechanisms to ensure safety but still there have flaws. The main concerned area is user authentication. For authentication, various biometric applications are used but once authentication is done in the begging there was no guarantee that the computer system is used by the authentic user or not. The intrusion detection system (IDS) is a particular procedure that is used to identify intruders by analyzing user behavior in the system after the user logged in. Host-based IDS monitors user behavior in the computer and identify user suspicious behavior as an intrusion or normal behavior. This paper discusses how an expert system detects intrusions using a set of rules as a pattern recognized engine. We propose a PIDE (Pattern Based Intrusion Detection) model, which is verified previously implemented SBID (Statistical Based Intrusion Detection) model. Experiment results indicate that integration of SBID and PBID approach provides an extensive system to detect intrusion.

Avoiding security vulnerabilities is very important for embedded systems. Dynamic Information Flow Tracking (DIFT) is a powerful technique to analyze SW with respect to security policies in order to protect the system against a broad range of security related exploits. However, existing DIFT approaches either do not exist for Virtual Prototypes (VPs) or fail to model complex hardware/software interactions.In this paper, we present a novel approach that enables early and accurate DIFT of binaries targeting embedded systems with custom peripherals. Leveraging the SystemC framework, our DIFT engine tracks accurate data flow information alongside the program execution to detect violations of security policies at run-time. We demonstrate the effectiveness and applicability of our approach by extensive experiments.

Deep Packet Inspection (DPI) is an essential building block implementing various services on data plane [5]. Usually, DPI engines are centered around efficient implementation of regular expressions both from the required memory and lookup time perspectives. In this paper, we explore and generalize original approaches used for packet classifiers [7] to regular expressions. Our preliminary results establish a promising direction for the efficient implementation of DPI engines.

More than 75% of the Internet traffic is now encrypted, while this percentage is constantly increasing. The majority of communications are secured using common encryption protocols such as SSL/TLS and IPsec to ensure security and protect the privacy of Internet users. Yet, encryption can be exploited to hide malicious activities. Traditionally, network traffic inspection is based on techniques like deep packet inspection (DPI). Common applications for DPI include but are not limited to firewalls, intrusion detection and prevention systems, L7 filtering and packet forwarding. The core functionality of such DPI implementations is based on pattern matching that enables searching for specific strings or regular expressions inside the packet contents. With the widespread adoption of network encryption though, DPI tools that rely on packet payload content are becoming less effective, demanding the development of more sophisticated techniques in order to adapt to current network encryption trends. In this work, we present HeaderHunter, a fast signature-based intrusion detection system even in encrypted network traffic. We generate signatures using only network packet metadata extracted from packet headers. Also, to cope with the ever increasing network speeds, we accelerate the inner computations of our proposed system using off-the-shelf GPUs.

Security and Privacy challenges are the most obstacles for the advancement of cloud computing and the erosion of trust boundaries already happening in organizations is amplified and accelerated by this emerging technology. Policy Management Frameworks are the most proper solutions to create dedicated security levels based on the sensitivity of resources and according to the mapping process between requirements cloud customers and capabilities of service providers. The most concerning issue in these frameworks is the rate of perfect matches between capabilities and requirements. In this paper, a reliable ring analysis engine has been introduced to efficiently map the security requirements of cloud customers to the capabilities of service provider and to enhance the rate of perfect matches between them for establishment of different security levels in clouds. In the suggested model a structural index has been introduced to receive the requirement and efficiently map them to the most proper security mechanism of the service provider. Our results show that this index-based engine enhances the rate of perfect matches considerably and decreases the detected conflicts in syntactic and semantic analysis.

As AI systems become more ubiquitous, securing them becomes an emerging challenge. Over the years, with the surge in online social media use and the data available for analysis, AI systems have been built to extract, represent and use this information. The credibility of this information extracted from open sources, however, can often be questionable. Malicious or incorrect information can cause a loss of money, reputation, and resources; and in certain situations, pose a threat to human life. In this paper, we use an ensembled semi-supervised approach to determine the credibility of Reddit posts by estimating their reputation score to ensure the validity of information ingested by AI systems. We demonstrate our approach in the cybersecurity domain, where security analysts utilize these systems to determine possible threats by analyzing the data scattered on social media websites, forums, blogs, etc.

With the popularity of the Internet and mobile intelligent terminals, the number of mobile applications is exploding. Mobile intelligent terminals trend to be the mainstream way of people's work and daily life online in place of PC terminals. Mobile application system brings some security problems inevitably while it provides convenience for people, and becomes a main target of hackers. Therefore, it is imminent to strengthen the security detection of mobile applications. This paper divides mobile application security detection into client security detection and server security detection. We propose a combining static and dynamic security detection method to detect client-side. We provide a method to get network information of server by capturing and analyzing mobile application traffic, and propose a fuzzy testing method based on HTTP protocol to detect server-side security vulnerabilities. Finally, on the basis of this, an automated platform for security detection of mobile application system is developed. Experiments show that the platform can detect the vulnerabilities of mobile application client and server effectively, and realize the automation of mobile application security detection. It can also reduce the cost of mobile security detection and enhance the security of mobile applications.

We propose a new architecture introducing AI to span the control layer and the data layer in SDON. This demonstration shows the cooperation of the AI engines in two layers in dealing with failure disposal.

Tactical edge networks represent a uniquely challenging environment from the communications perspective, due to their limited bandwidth and high node mobility. Several middleware communication solutions have been proposed to address those issues, adopting an evolutionary design approach that requires facing quite a few complications to provide applications with a suited network programming model while building on top of the TCP/IP stack. Information Centric Networking (ICN), instead, represents a revolutionary, clean slate approach that aims at replacing the entire TCP/IP stack with a new communication paradigm, better suited to cope with fluctuating channel conditions and network disruptions. This paper, stemmed from research conducted within NATO IST-161 RTG, investigates the effectiveness of Named Data Networking (NDN), the de facto standard implementation of ICN, in the context of tactical edge networks and its potential for adoption. We evaluated an NDN-based Blue Force Tracking (BFT) dissemination application within the Anglova scenario emulation environment, and found that NDN obtained better-than-expected results in terms of delivery ratio and latency, at the expense of a relatively high bandwidth consumption.

Detection of network attacks is the first step to network security. Many different methods for attack detection were proposed in the past. However, descriptions of these methods are often not complete and it is difficult to verify that the actual implementation matches the description. In this demo paper, we propose to use Complex Event Processing (CEP) for developing detection methods based on network flows. By writing the detection methods in an Event Processing Language (EPL), we can address the above-mentioned problems. The SQL-like syntax of most EPLs is easily readable so the detection method is self-documented. Moreover, it is directly executable in the CEP system, which eliminates inconsistencies between documentation and implementation. The demo will show a running example of a multi-stage HTTP brute force attack detection using Esper and its EPL.