Abstract. Assurance technologies for computer security have failed to have significant impacts in the marketplace, with the result that most of the computers connected to the internet are vulnerable to attack. This paper looks at the problem of malicious users from both a historical and practical standpoint. It traces the history of intrusion and intrusion detection from the early 1970s to the present day, beginning with a historical overview. The paper describes the two primary intrusion detection techniques, anomaly detection and signature-based misuse detection, in some detail and describes a number of contemporary research and commercial intrusion detection systems. It ends with a brief discussion of the problems associated with evaluating intrusion detection systems and a discussion of the difficulties associated with making further progress in the field. With respect to the latter, it notes that, like many fields, intrusion detection has been based on a combination of intuition and brute-force techniques. We suspect that these have carried the field as far as they can and that further significant progress will depend on the development of an underlying theoretical basis for the field.

Much of the computer security information regularly gathered and disseminated by individuals and organizations cannot currently be combined or compared because a "common language" has yet to emerge in the field of computer security. A common language consists of terms and taxonomies (principles of classification) which enable the gathering, exchange and comparison of information. This paper presents the results of a project to develop such a common language for computer security incidents. This project results from cooperation between the Security and Networking Research Group at the Sandia National Laboratories, Livermore, CA, and the CERT Coordination Center at Carnegie Mellon University, Pittsburgh, PA. This Common Language Project was not an effort to develop a comprehensive dictionary of terms used in the field of computer security. Instead, we developed a minimum set of "high-level" terms, along with a structure indicating their relationship (a taxonomy), which can be used to c...

Abstract not found

Conventional host-based and network-based intrusion and misuse detection systems have concentrated on detecting network-based and internal attacks, but little work has addressed host-based detection of low-level network attacks. A major reason for this is the misuse detection system’s dependence on audit data and the absence of low-level network data in audit trails. This work defines low-level IP vulnerabilities and distinguishes between low-level IP and IP-based vulnerabilities. Furthermore, we analyze a number of different low-level IP attacks and the vulnerabilities that they exploit. We develop attack signatures for each attack, and based upon our analysis, we determine a baseline collection of information needed to detect the attacks. We suggest locations within protocol stacks where the needed data can be collected. Finally, we generalize from the baseline audit data to try to predict audit content suitable not only for detecting these attacks, but possible future ones. 1

Distributed denial of service (DDoS) is a major threat to the availability of Internet services. The anonymity allowed by IP networking, together with the distributed, large scale nature of the Internet, makes DDoS attacks stealthy and difficult to counter. To make the problem worse, attack traffic is often indistinguishable from normal traffic. As various attack tools become widely available and require minimum knowledge to operate, automated anti-DDoS systems become increasingly important. Many current solutions are either excessively expensive or require universal deployment across many administrative domains. This paper proposes two perimeter-based defense mechanisms for Internet service providers (ISPs) to provide the anti-DDoS service to their customers. These mechanisms rely completely on the edge routers to cooperatively identify the flooding sources and establish rate-limit filters to block the attack traffic. The system does not require any support from routers outside or inside of the ISP, which not only makes it locally deployable but also avoids the stress on the ISP core routers. We also study a new problem of perimeter-based IP traceback and provide three solutions. We demonstrate analytically and by simulations that the proposed defense mechanisms react quickly in blocking attack traffic while achieving high survival ratio for legitimate traffic. Even when 40% of all customer networks attack, the survival ratio for traffic from the other customer networks is still close to 100%.

We describe an methodology for testing a software system for possible security flaws. Based on the observation that most security flaws are caused by the program's inappropriate interactions with the environment, and triggered by user's malicious perturbation on the environment (which we call an environment fault), we view the security testing problem as the problem of testing for the fault-tolerance properties of a software system. We consider each environment perturbation as a fault and the resulting security compromise a failure in the toleration of such faults. Our approach is based on the well known technique of fault-injection. Environment faults are injected into the system under test and system behavior observed. The failure to tolerate faults is an indicator of a potential security flaw in the system. An Environment-Application Interaction (EAI) fault model is proposed which guides us to decide what faults to inject. Based on EAI, we have developed a security testing methodology, and apply it to several applications. We successfully identified a number of vulnerabilities include vulnerabilities in Windows NT operating system.

Firewalls protect a trusted network from an untrusted network by filtering traffic according to a specified security policy. A diverse set of firewalls is being used today. As it is infeasible to examine and test each firewall for all possible potential problems, a taxonomy is needed to understand firewall vulnerabilities in the context of firewall operations. This paper describes a novel methodology for analyzing vulnerabilities in Internet firewalls. A firewall vulnerability is defined as an error made during firewall design, implementation, or configuration, that can be exploited to attack the trusted network that the firewall is supposed to protect. We examine firewall internals, and cross reference each firewall operation with causes and effects of weaknesses in that operation, analyzing twenty reported problems with available firewalls. The result of our analysis is a set of matrices that illustrate the distribution of firewall vulnerability causes and effects over firewall operations. These matrices are useful in avoiding and detecting unforeseen problems during both firewall implementation and firewall testing. Two case studies of Firewall-1 and Raptor illustrate our methodology.

Traditionally, computer security monitoring systems are built around the audit systems supplied by operating systems. These OS audit sources were not necessarily designed to meet modern security needs. This dissertation addresses this situation by categorizing monitoring systems based on their goals of detection and the time constraints of operation. This categorization is used to clarify what information is needed to perform detection as well as how the audit system should be structured to supply it in an appropriate manner. A prototype audit source was designed and constructed based on the information from the categorization. This audit system supplies information based on the type of detection to be performed. The new audit source was compared against an existing OS audit source and shown to have less overhead in many instances, generate a smaller volume of data, and generate useful information not currently available.

Software maintainers and auditors would benefit from a tool to help them focus their attention on functions that are likely to be the source of security vulnerabilities. However, the existence of such a tool is predicated on the ability to characterize a function's `security vulnerability likelihood.' Our hypothesis is that functions near a source of input are most likely to contain a security vulnerability. These functions should be a small percentage of the total number of functions in the system. To validate this hypothesis, we performed an experiment involving thirty one vulnerabilities in open source software. This paper describes the experiment, its outcome, and the tools used to conduct it. It also describes the FLF Finder, which is a tool that was developed using knowledge gathered from the outcome of the experiment. This tool automates the detection of high risk functions. To demonstrate the effectiveness of the FLF Finder, three open source applications with known vulnerabilities were tested. In addition to this test, a case study was performed on the privilege separation code in the OpenSSH server daemon.

When vulnerabilities are discovered in software, which often happens after deployment, they must be addressed as part of ongoing software maintenance. A mature software development organization should analyze vulnerabilities in order to determine how they, and similar vulnerabilities, can be prevented in the future. In this paper we present a structured method for analyzing and documenting the causes of software vulnerabilities. Applied during software maintenance, the method generates the information needed for improving the software development process, to prevent similar vulnerabilities in future releases. Our approach is based on vulnerability cause graphs, a structured representation of causes of software vulnerabilities.