Several analysis methods can be employed by an IDS, each with its own strengths and weaknesses, and their applicability to any given situation should be carefully considered.
There are two basic IDS analysis methods:
1. Pattern Matching (also called signature analysis), and
2. Anomaly detection
PATTERN MATCHING
Some of the first IDS products used signature analysis as their detection method and simply looked for known characteristics of an attack (such as specific packet sequences or text in the data stream) to produce an alert if that pattern was detected. If a new or different attack vector is used, it will not match a known signature and, thus, slip past the IDS.
ANOMALY DETECTION
Alternately, anomaly detection uses behavioral characteristics of a system’s operation or network traffic to draw conclusions on whether the traffic represents a risk to the network or host. Anomalies may include but are not limited to:
Multiple failed log-on attempts
Users logging in at strange hours
Unexplained changes to system clocks
Unusual error messages
Unexplained system shutdowns or restarts
Attempts to access restricted files
An anomaly-based IDS tends to produce more data because anything outside of the expected behavior is reported. Thus, they tend to report more false positives as expected behavior patterns change. An advantage to anomaly-based IDS is that, because they are based on behavior identification and not specific patterns of traffic, they are often able to detect new attacks that may be overlooked by a signature-based system. Often information from an anomaly-based IDS may be used to create a pattern for a signature-based IDS.
Host Based Intrusion Detection (HIDS)
HIDS is the implementation of IDS capabilities at the host level. Its most significant difference from NIDS is that related processes are limited to the boundaries of a single-host system. However, this presents advantages in effectively detecting objectionable activities because the IDS process is running directly on the host system, not just observing it from the network. This offers unfettered access to system logs, processes, system information, and device information, and virtually eliminates limits associated with encryption. The level of integration represented by HIDS increases the level of visibility and control at the disposal of the HIDS application.
Network Based Intrustion Detection (NIDS)
NIDS are usually incorporated into the network in a passive architecture, taking advantage of promiscuous mode access to the network. This means that it has visibility into every packet traversing the network segment. This allows the system to inspect packets and monitor sessions without impacting the network or the systems and applications utilizing the network.
Below you have other ways that instrusion detection can be performed:
Stateful Matching Intrusion Detection
Stateful matching takes pattern matching to the next level. It scans for attack signatures in the context of a stream of traffic or overall system behavior rather than the individual packets or discrete system activities. For example, an attacker may use a tool that sends a volley of valid packets to a targeted system. Because all the packets are valid, pattern matching is nearly useless. However, the fact that a large volume of the packets was seen may, itself, represent a known or potential attack pattern. To evade attack, then, the attacker may send the packets from multiple locations with long wait periods between each transmission to either confuse the signature detection system or exhaust its session timing window. If the IDS service is tuned to record and analyze traffic over a long period of time it may detect such an attack. Because stateful matching also uses signatures, it too must be updated regularly and, thus, has some of the same limitations as pattern matching.
Statistical Anomaly-Based Intrusion Detection
The statistical anomaly-based IDS analyzes event data by comparing it to typical, known, or predicted traffic profiles in an effort to find potential security breaches. It attempts to identify suspicious behavior by analyzing event data and identifying patterns of entries that deviate from a predicted norm. This type of detection method can be very effective and, at a very high level, begins to take on characteristics seen in IPS by establishing an expected baseline of behavior and acting on divergence from that baseline. However, there are some potential issues that may surface with a statistical IDS. Tuning the IDS can be challenging and, if not performed regularly, the system will be prone to false positives. Also, the definition of normal traffic can be open to interpretation and does not preclude an attacker from using normal activities to penetrate systems. Additionally, in a large, complex, dynamic corporate environment, it can be difficult, if not impossible, to clearly define “normal” traffic. The value of statistical analysis is that the system has the potential to detect previously unknown attacks. This is a huge departure from the limitation of matching previously known signatures. Therefore, when combined with signature matching technology, the statistical anomaly-based IDS can be very effective.
Protocol Anomaly-Based Intrusion Detection
A protocol anomaly-based IDS identifies any unacceptable deviation from expected behavior based on known network protocols. For example, if the IDS is monitoring an HTTP session and the traffic contains attributes that deviate from established HTTP session protocol standards, the IDS may view that as a malicious attempt to manipulate the protocol, penetrate a firewall, or exploit a vulnerability. The value of this method is directly related to the use of well-known or well-defined protocols within an environment. If an organization primarily uses well-known protocols (such as HTTP, FTP, or telnet) this can be an effective method of performing intrusion detection. In the face of custom or nonstandard protocols, however, the system will have more difficulty or be completely unable to determine the proper packet format. Interestingly, this type of method is prone to the same challenges faced by signature-based IDSs. For example, specific protocol analysis modules may have to be added or customized to deal with unique or new protocols or unusual use of standard protocols. Nevertheless, having an IDS that is intimately aware of valid protocol use can be very powerful when an organization employs standard implementations of common protocols.
Traffic Anomaly-Based Intrusion
Detection A traffic anomaly-based IDS identifies any unacceptable deviation from expected behavior based on actual traffic structure. When a session is established between systems, there is typically an expected pattern and behavior to the traffic transmitted in that session. That traffic can be compared to expected traffic conduct based on the understandings of traditional system interaction for that type of connection. Like the other types of anomaly-based IDS, traffic anomaly-based IDS relies on the ability to establish “normal” patterns of traffic and expected modes of behavior in systems, networks, and applications. In a highly dynamic environment it may be difficult, if not impossible, to clearly define these parameters.
Reference(s) used for this question:
Hernandez CISSP, Steven (2012-12-21). Official (ISC)2 Guide to the CISSP CBK, Third Edition ((ISC)2 Press) (Kindle Locations 3664-3686). Auerbach Publications. Kindle Edition.
and
Hernandez CISSP, Steven (2012-12-21). Official (ISC)2 Guide to the CISSP CBK, Third Edition ((ISC)2 Press) (Kindle Locations 3711-3734). Auerbach Publications. Kindle Edition.
and
Hernandez CISSP, Steven (2012-12-21). Official (ISC)2 Guide to the CISSP CBK, Third Edition ((ISC)2 Press) (Kindle Locations 3694-3711). Auerbach Publications. Kindle Edition.
Submit