5th Global Conference on Computing & Media Technology (2021) 1 Analysis of a Network Forensic Investigation Technique Shaina Ferdous1 and Dr. Vinesha Selvarajah3 Asia Pacific University of Innovation & Technology (APU), Malaysia 1 tp053500@mail.apu.edu.my, 2vinesha@apu.edu.my 1,2 Abstract— This paper discusses the topic of Network Forensics, the issues of Network Forensic techniques, and dives deeper into a specific Network Forensic Investigation model. The impact that Network Forensics has on current Digital Forensic trends and the challenges faced is also covered. Every step of the investigation model is thoroughly analyzed and based on the analysis a new investigation model is proposed that could also be used in its stead. The framework under analysis can work alongside other Network Forensic tools and techniques, making it quite versatile. Adaptability is crucial for digital forensic investigation models to fit into several different scenarios and instances and this framework has that part covered. There are no major requirements to utilize either of the investigative frameworks however since the proposed framework was derived from the investigative framework discussed earlier on in this paper, there is still room for improvements in the future for both investigative models. Introduction There is one thing that the current pandemic has taught us which is that anyone can become a victim to cyberattacks. This can be further proven by statistics showing that there has been an increase of 800% in overall cyberattacks in 2020 [1]. Businesses have been facing a large surge of cyber threats, especially ransomware attacks, now that everyone has started to work from their home offices rather than the company building itself. While it is being done to minimize contact people have with one another, it is putting the attacker at a much larger advantage. Each company has a dedicated team that makes sure the office network is as secure as it can be and harder to penetrate by most cyberattacks and so when an employee has to work from their office workstation, they do not have to worry about such things. All they need to focus on are their tasks for the day or week. But since the majority of people are encouraged to work from home due to the current pandemic, there is no one to look after your home network and so it automatically makes you more vulnerable to cyberattacks. This is what cybercriminals are taking advantage of and is also one of the top reasons why there is such an increase in cyberattacks within the past year. 1.1 Network Forensics Network Forensics is a subdivision of Digital Forensics which is a branch of Forensic Science [1]. They are all somehow related and so it is important to focus on all aspects of the Forensic Sciences. With that being said, the importance of Network Forensics is at its peak as it is the part of Digital Forensics that focuses on the examination of technology and data to understand the type of evidence that has been collected and how it is associated with the crime and assist with the investigation [1]. With this new information, professionals are also able to build a much stronger security system that will prevent such types of attacks from happening in the future. From the name, it is quite obvious that law enforcement will put more emphasis on network traffic data from the suspected network that was used for the cyberattack [1]. Some key information to look for include any form of communication between individuals, use of any certain keywords, network events logged, and also manipulations of existing files or folders [1]. Using this gathered information, it becomes much easier for cybercrime investigators to create a timeline of events that might have occurred. This is crucial in tracking down the source of the attack and also other security-related instances. 1.2 The Gap in Network Forensic Investigation Techniques The job of a forensic investigator consists of discovering how a crime took place, identifying the people involved, and creating a timeline of all the events that occurred which led to the crime [7]. With the plethora of techniques developed and used to assist in such activities, there is always one major issue with most of them which is the fact that information collected is always very abstract and avoids a lot of the technical details [7]. This leads to the issue of collecting unnecessary evidence just because of the “what if this is somehow related” concern most investigators often suffer from. Going through the piles of evidence then becomes an extremely strenuous and challenging task because now they have more data to filter through [7]. It is almost like a chain reaction. Another quite complex issue garnered from such techniques and also a huge issue of Network Forensics is that almost all the data gathered is in its rawest form which makes it even more difficult for humans to comprehend and decipher [7]. In general, Network Forensic investigations tend to involve a large amount of data that requires analysis which can be quite a tedious task to accomplish. Impact of Network Forensics Cyber Security can be simply defined as the act of protecting electronic devices and data from attacks of ill intent. The involvement of Forensic Investigations assists with collecting evidence to catch the attackers. Multiple methods have been constructed over the years to combat such malicious attacks and Network Forensics has played a huge role in it by broadening the horizons and opportunities for Digital Forensic Investigations. 5th Global Conference on Computing & Media Technology (2021) 2.1 Detection and Prevention One of the first steps for any Digital Forensic Investigation is the detection of the attack and with the help of Network Forensics, it has now become much easier to do since it allows us to monitor the network traffic of attack hotspots and analyze the attack using tools like Wireshark, Splunk, Snort, Network Miner, etc. By analyzing the attack, you can then take further precautions on how to prevent such attacks from happening in the future since you can predict the attack patterns through Network Forensic tools and techniques. 2.2 Simplified and Added Security The implementation of Network Forensic techniques has not only simplified the investigation of attacks but also the identification process of weaknesses within a network has become an extra layer of security for most. Network Forensics has proved to us multiple times that having weaknesses or vulnerabilities within one’s system or network can be used to one’s advantage. Now that it has been established what could potentially be exploited by a cybercriminal in the future, we can work on rectifying this vulnerability or even use it to create a honeypot. Luring attackers into a situation where they think they are in control, even if that is not the case at all, is defined as a honeypot and is a common tactic used by law enforcement to catch cybercriminals. This just shows that Network Forensics is quite versatile and has changed the game for both cyber professionals and law enforcement. 2.3 Troubleshooting Performance Issues When working in a corporate environment, often there will be scenarios where the network is facing some performance issues. With the help of Network Forensics, you are not only able to identify the problem but also troubleshoot the issue into optimizing the performance quality of the network which goes to show that there is a need for Network Forensics, not only in cybercrime investigations but also in the corporate world which further proves its versatility. 2.4 Identify Source of Data Leaks Data leaks are a very common trend lately as with the advancement of technology, data is becoming just as valuable as money. Whether it be an outsider trying to gain access or an insider that already has access to the data but would like to leak it for ulterior motives, there is always a need for data protection and monitoring especially with the recent rise in ransomware attacks. Since all information is transferred through a network, it makes sense that Network Forensics would also play a key role in this and assist with finding out information like where the leak occurred, how it occurred, and who might be responsible for it. A Review of Network Forensic Technique This section discusses the Multi-layer Multi-model Traffic Analysis (MMTA) forensic technique which is commonly used in the industry because of its attention to detail which is missed by most other Network Forensic techniques. The focus of the area in this investigation model is not only the network itself but also servers and other assets and devices the information is 2 flowing through. Analysis of abnormal behavioural patterns and how this model correlates with the data gathered is what piqued my interest in it. 3.1 Multi-layer Multi-model Traffic Analysis (MMTA) All devices that are exposed to the internet are at risk of being compromised and attacked especially in this day and age where technology just keeps advancing at an exponential rate. Advancement in technology calls for advancement in forensic investigation methodologies with various areas of focus, especially in Network Forensics. Since traffic analysis of a network for a very common method of approaching the topic of Network Forensics, Dinil Mon Divakaran decided to develop the Multi-layer Multi-model Traffic Analysis (MMTA) model, in 2017, which consists of three main phases (Divankaran, 2017). The diagram below is a depiction of the investigative model. Fig. 1 MMTA framework [5] Phase 1 is modelling and analyzing network sessions to detect anomalous or suspicious patterns [5]. This is the phase where the network in question is thoroughly examined. Every TCP connection is monitored, every session created by other devices is analyzed, every single movement within the traffic of the network is tracked. Phase 2 is detecting scans and illegitimate TCP state sequences [5]. The first step before performing any attack is to scan the networks and servers used by the target to find any open ports that can be exploited to find out the services running in the target machine as well as the versions. This is where security breaches begin [5]. In the stage of the analysis, instead of only detecting scans, illegitimate TCP state sequences are observed. Phase 3 is evidence correlation and decision making [5]. In this phase, based on all the evidence gathered from phase 1 and phase 2, it can be revealed what sort of attack occurred or what might have happened and how the attacker gained access to the system from the network. If the attack is still active then methods of stopping the attack need to be implemented and if it has already been affected then future precautions must be taken. 5th Global Conference on Computing & Media Technology (2021) This can lead to how this type of attack can be prevented from happening again in the future. Multi-Layer Multi-model Traffic Analysis 4.1 Phase 1: Model, Analyze, and Detect Anomalous Patterns As mentioned earlier the MMTA framework consists of three major phases which were briefly discussed in the previous section. Now the details of each phase will be elaborated on. Before we begin, we must first establish what exactly is a flow and a session. A flow is a set of packets consisting of five tuples which are the source and destination ports, source and destination IP addresses, and protocol [5]. If two flows have the same five tuples but have a period of inactivity between them then they are considered to be two separate flows [5]. So, with this theory, every TCP connection is a different flow. Based on the arrival time between packets, inter-arrival time, it can be determined whether a set of flows can be considered a session or not [5]. A session needs to have the inter-arrival time between flows to be less than a certain value [5]. Hence, if there is a sufficient period of inactivity detected between two similar sessions, then they are considered to be two separate sessions. Often, it can be a little more efficient to filter through the sessions by analyzing three tuples instead of five so the main focus would be on the destination port, destination IP address, and protocol [5]. In this way, a session can be sifted through by identifying the same IP addresses. This is a common way to analyze traffic from a popular website but in the case of bots or DDOS attacks, it would be more helpful to look for the source IP address rather than the destination IP address [5]. And so, these are some ways to complete the first step of the MMTA framework. 3 This entire process starts when the client does an active open which is when the client creates a connection to a listening port. It is followed by a SYN_RCVD from the server which tells us that the server has received a synchronized request and is currently waiting for an acknowledgement from the client. The client then acknowledges the request from the server and they establish a connection between the client and the server. Once all the necessary activity between the client and server is done, the client then proceeds to actively close the connection which the server follows with a passive close. A FIN is sent to the client along with an acknowledgement request for that FIN request and the server waits for it to be acknowledged by the client for the connection to be completely closed. The next diagram which is called the TCP state transition diagram which is a little more intricate than the previous TCP state sequence diagram but applies the same concept. 4.2 Phase 2: Detect Scans and Illegitimate TCP Sequences Moving on to the next phase of detecting and scanning illegitimate TCP state sequence behaviour, we must first understand the regular TCP states of establishing and terminating a connection between the client and the server. This is well depicted in the diagram below Fig. 3 TCP State Transition Diagram [4] Fig. 2 TCP State Sequence [4] The hard black arrow represents the normal client transition behaviour and the dashed arrow represents the server transitions. Connection establishment between the client and the server begins at the CLOSED state at the top of the diagram and the arrows facing the ESTABLISHED state show the opening of a connection while the arrows leaving the ESTABLISHED state show the closing of that connection. The ESTABLISHED state represents an open connection and this is where all the data transferring happens between the client and the server. On the bottom left of the diagram inside the dashed box labelled active close, this is showing the states of the client trying to initiate the closing of a connection to the server. The dashed box on the bottom right labelled passive close is the state of the server acknowledging the connection closing request of 5th Global Conference on Computing & Media Technology (2021) the client and eventually leading to the CLOSED state back at the top where the connection has finally ended. These two diagrams must be something that all Network Forensic experts should as they can be very useful in determining whether there is any suspicious activity happening inside of a network. Phase 2 of the MMTA framework is all about the TCP states and the behavioural analysis of these states. This is also where most of the illegitimate TCP state sequences can be found and investigated further and so is a very crucial part of the MMTA framework. 4 how to use the tool itself to gather data. Once the data on the network traffic has been gathered, the three phases of the MMTA framework can be applied to reconstruct the attack and derive a conclusion based on the evidence gathered. Challenges Faced With any new technique or tool developed, there will always be advantages and disadvantages to them and challenges will be faced both in the development process as well as in the usage of the tool and technique. The following are some of the challenges encountered during the use of the MMTA framework and also in the field of Network Forensics. 4.3 Phase 3: Evidence Correlation and Decision Making The final stage of the MMTA framework, the evidence correlation phase, is where the relationship between the anomalous patterns detected in phase 1 and the illegitimate TCP state behaviour in phase 2 can be put together to derive a conclusion on the attack. There are few different methods to proceed in this phase, a common one being the in-depth analysis of IP addresses as they can be targeted by time and destination. Often, the same IP addresses indicate anomalous behaviour which could lead to illegitimate flows and so it is best to start your investigation from there [5]. Although the time of the flows recorded can be utilized, not all related anomalies occur within the same time frame [5]. They can be hours or even days apart. On the other hand, it can also be helpful to analyze the Round Trip Times (RTTs) of the packets involved and find similarities between those packets [5]. Correlating information based on IP addresses is a much safer and reliable option as it produces optimal results. For example, if a bot scans multiple machines before selecting one to establish a session with its master, there would be two sessions with the same source IP address which would make it easier to identify the suspicious sessions to analyze [5]. General Requirements The use of this framework is not defined by any certain task that must be completed to be able to use the MMTA framework. This implies that there are no specific requirements that need to be fulfilled for the MMTA framework to be utilized. With that being said, this investigational methodology can be implemented on any type of network forensic tool. Some of the most popular and commonly used network forensic tools in the current market are the following: • Network Miner • Wireshark • Tcpdump • Splunk • Snort The only basic requirement that is quite crucial to have is the understanding of the different states of TCP sessions between the client and server because that is what this entire investigational framework is mostly dependent on. Being well versed with an intensive network forensic tool can be quite handy as well since not much time will be wasted in learning 6.1 Detection of False Positives During the use of the MMTA framework, there have been multiple cases of detecting false positives in the first and second phase of evidence collection and anomalous pattern detection [5]. False-positive is defined as when a test result turns out to be positive, even if that is not the case. In this scenario, the detection of a false positive would imply that there has been some suspicious activity, or an anomalous pattern or an illegitimate session has been discovered [5]. Even if that is not the case, this would hinder the final stage of the evidence correlation and decision-making process and could lead the investigation and network traffic analysis process in the wrong direction. Hence, it is important to be cautious of these types of results and perform a thorough analysis of the evidence gathered to avoid such scenarios as a lot of manpower and resources could be affected by this. 6.2 Data Lost After Transmission Network Forensics can be a bit more tricky than common Digital Forensics practices as there is a possibility of the data carried out within a network being lost after transmission, which is usually not the case for the latter. Surprisingly, this happens quite often and most of the time the cause of data disappearing is unknown. This provides an extra challenge for most Network Forensic Investigators to hunt for evidence and dig deeper for information that has the potential to be valuable in finding the culprit in an attack. 6.3 Privacy Laws There is also the existence of data protection and privacy laws that might restrict some network forensic investigational methods and tools [1]. Hence, before using a network forensic tool, it is always best to make sure it does not violate any privacy laws. If you are acting as a third-party investigator and have been hired by a company to analyze their network then it is always best to make sure that you have the right permissions and authorizations achieved before beginning the examination and analysis using your tool and methodology of choice. These are just some extra precautionary steps that people who are trying to pursue a career in the digital forensic and network forensic field should be aware of. 5th Global Conference on Computing & Media Technology (2021) Critical Discussion of Analysis Process There are two ways to approach the art of Network Forensics, the first being “catch it as you can” and the second being “stop, look and listen” [6]. With the first method, the investigator captures all network traffic for further analysis and is considered to be quite a time-consuming process and requires a lot of storage space [6]. The second method allows the investigator to analyze and examine every single data packet flowing through the suspected network and only capture the data packets that seem suspicious and need further examination [6]. This method is equally time-consuming as the first and needs a lot of processing power but does not require as much storage space. Although the investigator does have the liberty to capture all network traffic by using the “catch it as you can” approach, the MMTA framework works better with the “stop, look and listen” approach. The reason for that being is that when conducting a forensic analysis on the traffic of a network, time is usually of the essence and it is important to be as efficient as possible as information passed on through packets can be easily lost over time. In case of a network that is facing heavy traffic due to multiple reasons, capturing all packet information and going through every packet one by one might turn out to be ineffective as a lot of the times these investigations are done on a hunch and there might not even be any malicious activity happening in that network. This also proves that the “catch as you can” approach is best suited for networks that are confirmed to have malicious activity happening by other sources or if the network traffic is not very heavy. When performing the first phase of the MMTA framework where the investigator needs to focus on finding anomalous patterns within the traffic, the process would move along much smoothly if only the suspicious-looking packets were captured to have a further examination done. In this way, a lot of space is saved but the issue of taking up a significant amount of processing power remains as scanning through each packet before capturing only the suspicious ones will require a much larger computational capacity. There is also the issue of false positives being detected using this method so it is very crucial to be thorough and make sure the right packets are being analyzed. It is quite obvious by now that both of these methods are in no way time-efficient and are quite tedious activities, however, that is just a part of any investigational process. Whether it be a physical crime investigation or a cybercrime investigation, the analysis and examination part of the investigation is always the most laborious, extensive, and slow. There have been several professionals who have studied the different existing investigational methodologies, techniques, and tools to find a faster way to accomplish and gather results. Quite a few of them have succeeded in shortening the entire investigational process based on the situation of the crime. However, the analysis and examination step is still the longest portion of the investigation because it is all part of the process and must be followed to gather the best results. 5 Standard Operating Procedure As someone who is trying to pursue a career in the field of digital forensics, it has become mandatory to know the steps of performing a forensic investigation by heart. The steps consist of Authorization, Identification, Evidence Collection, Analysis and Examination, Reconstruction, and lastly Documentation and Presentation. It has become evident over the years that all of these steps, although necessary, cannot be applied at every single digital crime scene and so several different types of methodologies based on various scenarios have been developed. This goes to show that even if those 6 steps can be referred to as a standard forensic investigation procedure, there is still room for change within those steps. With that being said, the Multi-layer Multi-model Traffic Analysis framework, however much versatile and effective in network forensic investigations, still has room for improvement and suggestion. The sole focus of this entire investigational model is on network traffic, which is very important in network forensics, however, other areas of the topic need to be brought to focus as well. By studying the TCP state sequences, a lot can be discovered about one’s network activity. Nevertheless, TCP is not the only network communication protocol out there that should be highlighted. Although it is the most frequently used protocol, the focus should be directed to UDP as well since it is also a protocol used in the transport layer of the OSI model. Other protocols like UDP, DHCP, DNS, FTP, HTTP, IMAP, POP, SMTP, and so on should also be highlighted based on the type of network activity being tracked. Each of these protocols has different uses like for example, SMTP, IMAP, and POP are used for sending emails and FTP is used for transferring files and so on [3]. 8.1 Proposed Network Forensic Technique Keeping in mind the discussion from the previous section, below are some steps that could be followed that could benefit and improve the current MMTA framework. Step 1 Gain Authorization This step can be very easily overlooked but could really determine whether the evidence collected is admissible in court and so it is crucial to gain the right legal authorizations before starting the investigation. The tools being used for the investigation should also be checked at this point and verified whether they follow privacy and data protection laws Step 2 Identification of Layer To my knowledge, this step was completely overlooked by the MMTA framework. This is the step where the attack is thorough to find out which layers of the OSI model should be targeted in the investigation. The OSI model has 7 different layers and the diagram below represents the process of communication between two network devices. 5th Global Conference on Computing & Media Technology (2021) 6 also a deep dive on TCP sequences and session analysis within network traffics which can be used in future investigations involving Network Forensics and be useful in other types of Digital Forensic Investigations. The framework proposed is quite basic as it currently focuses on collecting data from traffic and later doing analysis on it but with enhancements, the framework can be utilized on live traffic which is why both MMTA and proposed frameworks have room for improvement. Fig. 4 Data transfer from one device to another through OSI model [3] Usually, the main focus is given to the Transport, Network, and Data-link layers of the model. Once it has been identified which layer or layers should be focused on, we can move on to the next step. Step 3 Identification of Network Protocol Based on the layer selection from the previous step, the network protocol of focus can be decided upon. If for example, both the Network and Transport layers have been targeted, then it can be narrowed down to what protocols need more focus. Each protocol has its functionality so it can be easily determined which protocols to keep a lookout for once it has been established what exactly you need to be looking for. Step 4 Model, Analyze, Detect Anomalous Patterns (continuation of MMTA framework) Now that it has been identified which protocols to focus on most, it can narrow down the search and examination of packets and anomalous behaviour. Step 5 Detect Scans and Illegitimate Network Protocol Behavior Based on the sequences of the network protocol in the examination, it can be determined whether illegitimate activity or sequences are being sent across the network in question. References [1] Afifi-Sabet, K. (2021). What is network forensics? Retrieved May 2021, from https://www.itpro.co.uk/cyber-attacks/31660/what-isnetwork-forensics [2] Anon. (2021). Tcp State Transition Diagram. Retrieved May 2021, from https://www.fishercom.xyz/digitaltransmission/tcp-state-transitiondiagram.html#:~:text=A%20TCP%20connection%20goe s%20through,indicates%20associated%20events%20and %20actions.&text=Connection%20termination%20goes %20from%20the%20ESTABLISHED%20state%20to% 20 [3] Anon. (n.d.). Network protocols. Retrieved May 2021, from https://www.manageengine.com/networkmonitoring/network-protocols.html [4] Anon. (n.d.). TCP State Transition Diagram. Retrieved 2021, from https://flylib.com/books/en/3.223.1.188/1/ [5] Divankaran, D. M. (2017). Evidence gathering for network security and forensics. DFRWS 2017 Europe — Proceedings of the Fourth Annual DFRWS Europe, 20, S56-S65. [6] Kostadinov, D. (2020). Network Forensic Overview. Retrieved May 2021, from https://resources.infosecinstitute.com/topic/networkforensicsoverview/#:~:text=According%20to%20%E2%80%9CC omputer%20Forensics%3A%20Network,be%20used%2 0for%20network%20forensics. [7] Nishikant R. Khaire, 2. V. (2015). Recent Trends and Challenges Of Network Forensic. International Journal of Advanced Computational Engineering and Networking, ISSN: 2320-210, 3(12), 7. Step 6 Evidence Correlation and Decision Making In the final step, it is helpful to reconstruct the attack with the evidence collected and use that data to proceed with the decision making and respond to the attack. Conclusions In this paper, analysis was done on the MMTA Network Forensic investigation framework that was created for gathering evidence and detecting malicious activity or attacks in traffic sessions. Based on the analysis a standard operating procedure and a new investigative procedure was proposed. There was