Recovering and Examining Computer Forensic Evidence by Noblett et al. (Forensic Science Communications, October 2000)
October 2000 - Volume 2 - Number 4
Recovering and Examining Computer Forensic EvidenceMichael G. Noblett
Booz-Allen & Hamilton
Falls Church, Virginia
Mark M. Pollitt
Computer Analysis and Response Team
Federal Bureau of Investigation
Lawrence A. Presley
Forensic Science Training Unit
The world is becoming a smaller place in which to live and work. A technological revolution in communications and information exchange has taken place within business, industry, and our homes. America is substantially more invested in information processing and management than manufacturing goods, and this has affected our professional and personal lives. We bank and transfer money electronically, and we are much more likely to receive an E-mail than a letter. It is estimated that the worldwide Internet population is 349 million (CommerceNet Research Council 2000).
|In this information technology age, the needs of law enforcement are changing as well. Some traditional crimes, especially those concerning finance and commerce, continue to be upgraded technologically. Paper trails have become electronic trails. Crimes associated with the theft and manipulations of data are detected daily. Crimes of violence also are not immune to the effects of the information age. A serious and costly terrorist act could come from the Internet instead of a truck bomb. The diary of a serial killer may be recorded on a floppy disk or hard disk drive rather than on paper in a notebook.||
Just as the workforce has gradually converted from manufacturing goods to processing information, criminal activity has, to a large extent, also converted from a physical dimension, in which evidence and investigations are described in tangible terms, to a cyber dimension, in which evidence exists only electronically, and investigations are conducted online.
Computer forensic science was created to address the specific and articulated needs of law enforcement to make the most of this new form of electronic evidence. Computer forensic science is the science of acquiring, preserving, retrieving, and presenting data that has been processed electronically and stored on computer media. As a forensic discipline, nothing since DNA technology has had such a large potential effect on specific types of investigations and prosecutions as computer forensic science.
Computer forensic science is, at its core, different from most traditional forensic disciplines. The computer material that is examined and the techniques available to the examiner are products of a market-driven private sector. Furthermore, in contrast to traditional forensic analyses, there commonly is a requirement to perform computer examinations at virtually any physical location, not only in a controlled laboratory setting. Rather than producing interpretative conclusions, as in many forensic disciplines, computer forensic science produces direct information and data that may have significance in a case. This type of direct data collection has wide-ranging implications for both the relationship between the investigator and the forensic scientist and the work product of the forensic computer examination.
Computer forensic science is largely a response to a demand for service from the law enforcement community. As early as 1984, the FBI Laboratory and other law enforcement agencies began developing programs to examine computer evidence. To properly address the growing demands of investigators and prosecutors in a structured and programmatic manner, the FBI established the Computer Analysis and Response Team (CART) and charged it with the responsibility for computer analysis. Although CART is unique in the FBI, its functions and general organization are duplicated in many other law enforcement agencies in the United States and other countries.
An early problem addressed by law enforcement was identifying resources within the organization that could be used to examine computer evidence. These resources were often scattered throughout the agency. Today, there appears to be a trend toward moving these examinations to a laboratory environment. In 1995, a survey conducted by the U.S. Secret Service indicated that 48 percent of the agencies had computer forensic laboratories and that 68 percent of the computer evidence seized was forwarded to the experts in those laboratories. As encouraging as these statistics are for a controlled programmatic response to computer forensic needs, the same survey reported that 70 percent of these same law enforcement agencies were doing the work without a written procedures manual (Noblett 1995).
Computer forensic examinations are conducted in forensic laboratories, data processing departments, and in some cases, the detective’s squad room. The assignment of personnel to conduct these examinations is based often on available expertise, as well as departmental policy. Regardless of where the examinations are conducted, a valid and reliable forensic examination is required. This requirement recognizes no political, bureaucratic, technological, or jurisdictional boundaries.
There are ongoing efforts to develop examination standards and to provide structure to computer forensic examinations. As early as 1991, a group of six international law enforcement agencies met with several U.S. federal law enforcement agencies in Charleston, South Carolina, to discuss computer forensic science and the need for a standardized approach to examinations. In 1993, the FBI hosted an International Law Enforcement Conference on Computer Evidence that was attended by 70 representatives of various U.S. federal, state, and local law enforcement agencies and international law enforcement agencies. All agreed that standards for computer forensic science were lacking and needed. This conference again convened in Baltimore, Maryland, in 1995, Australia in 1996, and the Netherlands in 1997, and ultimately resulted in the formation of the International Organization on Computer Evidence. In addition, a Scientific Working Group on Digital Evidence (SWGDE) was formed to address these same issues among federal law enforcement agencies.
A New Relationship
Forensic science disciplines have affected countless criminal investigations dramatically and have provided compelling testimony in scores of trials. To enhance objectivity and to minimize the perception of bias, forensic science traditionally has remained at arms length from much of the actual investigation. It uses only those specific details from the investigation that are necessary for the examination. These details might include possible sources of contamination at the crime scene or fingerprints of individuals not related to the investigation who have touched the evidence. Forensic science relies on the ability of the scientists to produce a report based on the objective results of a scientific examination. The actual overall case may play a small part in the examination process. As a case in point, a DNA examination in a rape case can be conducted without knowledge of the victim’s name, the subject, or the specific circumstances of the crime.
Conversely, computer forensic science, to be effective, must be driven by information uncovered during the investigation. With the average storage capacity in a personally owned microcomputer approaching 30 gigabytes (GB; Fischer 1997), and systems readily available that have 60-GB storage capacity or more, it is likely to be impossible from a practical standpoint to completely and exhaustively examine every file stored on a seized computer system. In addition, because computers serve such wide and varied uses within an organization or household, there may be legal prohibitions against searching every file. Attorney or physician computers may contain not only evidence of fraud but probably also client and patient information that is privileged. Data centrally stored on a computer server may contain an incriminating E-mail prepared by the subject as well as E-mail of innocent third parties who would have a reasonable expectation of privacy.
As difficult as it would be to scan a directory of every file on a computer system, it would be equally difficult for law enforcement personnel to read and assimilate the amount of information contained within the files. For example, 12 GB of printed text data would create a stack of paper 24 stories high. For primarily pragmatic reasons, computer forensic science is used most effectively when only the most probative information and details of the investigation are provided to the forensic examiner. From this information, the examiner can create a list of key words to cull specific, probative, and case-related information from very large groups of files. Even though the examiner may have the legal right to search every file, time limitations and other judicial constraints may not permit it. The examination in most cases should be limited to only well-identified probative information.
Forensic science has historically produced results that have been judged to be both valid and reliable. For example, DNA analysis attempts to develop specific identifying information relative to an individual. To support their conclusions, forensic DNA scientists have gathered extensive statistical data on the DNA profiles from which they base their conclusions. Computer forensic science, by comparison, extracts or produces information. The purpose of the computer examination is to find information related to the case. To support the results of a computer forensic examination, procedures are needed to ensure that only the information exists on the computer storage media, unaltered by the examination process. Unlike forensic DNA analysis or other forensic disciplines, computer forensic science makes no interpretive statement as to the accuracy, reliability, or discriminating power of the actual data or information.
Beyond the forensic product and the case-related information needed to efficiently perform the work, there is another significant difference between most traditional forensic science and computer forensic science. Traditional forensic analysis can be controlled in the laboratory setting and can progress logically, incrementally, and in concert with widely accepted forensic practices. In comparison, computer forensic science is almost entirely technology and market driven, generally outside the laboratory setting, and the examinations present unique variations in almost every situation.
These dissimilarities aside, both the scientific conclusions of traditional forensic analyses and the information of computer forensic science are distinctive forensic examinations. They share all the legal and good laboratory practice requirements of traditional forensic sciences in general. They both will be presented in court in adversarial and sometimes very probing proceedings. Both must produce valid and reliable results from state-of-the-art procedures that are detailed, documented, and peer-reviewed and from protocols acceptable to the relevant scientific community (ASCLD/LAB 1994).
As laboratories begin to examine more computer-related evidence, they must establish policies regarding computer forensic examinations and, from these policies, develop protocols and procedures. The policies should reflect the broad, community-wide goal of providing valid and reproducible results, even though the submissions may come from diverse sources and present novel examination issues. As the laboratory moves from the policy statement to protocol development, each individual procedure must be well-documented and sufficiently robust to withstand challenges to both the results and methodology.
However, computer forensic science, unlike some of its traditional forensic counterparts, cannot rely on receiving similar evidence in every submission. For instance, DNA from any source, once cleared of contaminants and reduced to its elemental form, is generic. From that point, the protocols for forensic DNA analysis may be applied similarly to all submissions. The criminal justice system has come to expect a valid and reliable result using those DNA protocols. For the following reasons, computer forensic science can rarely expect these same elements of standardized repetitive testing in many of its submissions:
- Operating systems, which define what a computer is and how it works, vary among manufacturers. For example, techniques developed for a personal computer using the Disk Operating System (DOS) environment may not correspond to operating systems such as UNIX, which are multi-user environments.
- Applications programs are unique.
- Storage methods may be unique to both the device and the media.
Typical computer examinations must recognize the fast-changing and diverse world in which the computer forensic science examiner works.
Computer evidence represented by physical items such as chips, boards, central processing units, storage media, monitors, and printers can be described easily and correctly as a unique form of physical evidence. The logging, description, storage, and disposition of physical evidence are well understood. Forensic laboratories have detailed plans describing acceptable methods for handling physical evidence. To the extent that computer evidence has a physical component, it does not represent any particular challenge. However, the evidence, while stored in these physical items, is latent and exists only in a metaphysical electronic form. The result that is reported from the examination is the recovery of this latent information. Although forensic laboratories are very good at ensuring the integrity of the physical items in their control, computer forensics also requires methods to ensure the integrity of the information contained within those physical items. The challenge to computer forensic science is to develop methods and techniques that provide valid and reliable results while protecting the real evidence—the information—from harm.
To complicate the matter further, computer evidence almost never exists in isolation. It is a product of the data stored, the application used to create and store it, and the computer system that directed these activities. To a lesser extent, it is also a product of the software tools used in the laboratory to extract it.
Computer forensic science issues must also be addressed in the context of an emerging and rapidly changing environment. However, even as the environment changes, both national and international law enforcement agencies recognize the need for common technical approaches and are calling for standards (Pollitt 1998). Because of this, a model (see Figure 1) must be constructed that works on a long-term basis even when short-term changes are the rule rather than the exception. The model that we describe is a three-level hierarchical model consisting of the following:
- An overarching concept of the principles of examination,
- Policies and practices, and
- Procedures and techniques.
Principles of examinations are large-scale concepts that almost always apply to the examination. They are the consensus approaches as to what is important among professionals and laboratories conducting these examinations. They represent the collective technical practice and experience of forensic computer examiners.
Organizational policy and practices are structural guidance that applies to forensic examinations. These are designed to ensure quality and efficiency in the workplace. In computer forensic science, these are the good laboratory practices by which examinations are planned, performed, monitored, recorded, and reported to ensure the quality and integrity of the work product.
Procedures and techniques are software and hardware solutions to specific forensic problems. The procedures and techniques are detailed instructions for specific software packages as well as step-by-step instructions that describe the entire examination procedure (Pollitt 1995).
As an overall example, a laboratory may require that examinations be conducted, if possible and practical, on copies of the original evidence. This requirement is a principle of examination. It represents a logical approach taken by the computer forensic science community as a whole, and it is based on the tenet of protecting the original evidence from accidental or unintentional damage or alteration. This principle is predicated on the fact that digital evidence can be duplicated exactly to create a copy that is true and accurate.
Creating the copy and ensuring that it is true and accurate involves a subset of the principle, that is, policy and practice. Each agency and examiner must make a decision as to how to implement this principle on a case-by-case basis. Factors in that decision include the size of the data set, the method used to create it, and the media on which it resides. In some cases it may be sufficient to merely compare the size and creation dates of files listed in the copy to the original. In others, it may require the application of more technically robust and mathematical rigorous techniques such as a cyclical redundancy check (CRC) or calculating a message digest (MD).
CRC and MD are computer algorithms that produce unique mathematical representations of the data. They are calculated for both the original and the copy and then compared for identity. The selection of tools must be based on the character of the evidence rather than simply laboratory policy. It is likely that examiners will need several options available to them to perform this one function.
An examiner responsible for duplicating evidence must first decide an appropriate level of verification to weigh time constraints against large file types. The mathematical precision and discriminating power of these algorithms are usually directly proportional to the amount of time necessary to calculate them. If there were 1 million files to be duplicated, each less than 1 kilobyte in size, time and computational constraints would likely be a major determining factor. This circumstance would probably result in a decision to use a faster, but less precise and discriminating, data integrity algorithm.
Having decided how best to ensure the copy process will be complete and accurate, the next step is the actual task. This is a subset of the policy and practice, that is, procedures and techniques. These most closely represent the standard cookbook approach to protocol development. They are complete and contain required detailed steps that may be used to copy the data, verify that the operation was complete, and ensure that a true and accurate copy has been produced.
Again, as Figure 1 illustrates, a principle may spawn more that one policy, and those policies can accept many different techniques. The path an examiner takes in each case is well-documented and technologically sound for that particular case. It may not, however, be the same path the examiner takes with the next case. Traditional forensic examinations, such as the DNA examination of blood recovered from a crime scene, lend themselves to a routine and standardized series of steps that can be repeated in case after case. There is generally no such thing as generic computer evidence procedures. The evidence is likely to be significantly different every time a submission is received by the laboratory and will likely require an examination plan tailored to that particular evidence. Although this situation may present a recurrent consideration of management checks and controls within the laboratory setting, it is a consideration that must be addressed and improved if this emerging forensic discipline is to remain an effective and reliable tool in the criminal justice system.
Valid and reliable methods to recover data from computers seized as evidence in criminal investigations are becoming fundamental for law enforcement agencies worldwide. These methods must be technologically robust to ensure that all probative information is recovered. They must also be legally defensible to ensure that nothing in the original evidence was altered and that no data was added to or deleted from the original. The forensic discipline of acquiring, preserving, retrieving, and presenting data that has been processed electronically and stored on computer media is computer forensic science.
This article examined issues surrounding the need to develop laboratory protocols for computer forensic science that meet critical technological and legal goals. Computer forensic scientists need to develop ongoing relationships with the criminal justice agencies they serve. The reasons for these relationships include the following:
- In their efforts to minimize the amount of data that must be recovered and to make their examinations more efficient and effective, computer forensic scientists must have specific knowledge of investigative details. This is a clear requirement that is generally more demanding than traditional forensic science requests, and it places more reliance on case information.
- Courts are requiring that more information rather than equipment be seized. This requires cooperative efforts between law enforcement officers and the computer forensic scientist to ensure that the technical resources necessary for the execution of the search warrant are sufficient to address both the scope and complexity of the search.
- Computers may logically contain both information identified in the warrant as well as information that may be constitutionally protected. The computer forensic scientist is probably the most qualified person to advise both the investigator and prosecutor as to how to identify technical solutions to these intricate situations.
Developing computer examination protocols for forensic computer analysis is unique for several reasons:
- Unlike some traditional forensic analyses that attempt to gather as much information as possible from an evidence sample, computer forensic analysis attempts to recover only probative information from a large volume of generally heterogenous information.
- Computer forensic science must take into account the reality that computer forensic science is primarily market driven, and the science must adapt quickly to new products and innovations with valid and reliable examination and analysis techniques.
- The work product of computer forensic science examinations also differs from most traditional forensic work products. Traditional forensic science attempts to develop a series of accurate and reliable facts. For example, the DNA extracted from blood found at a crime scene can be matched to a specific person to establish the fact that the blood was shed by that person to the exclusion of all other individuals. Computer forensic science generally makes no interpretive statement as to the accuracy or reliability of the information obtained and normally renders only the information recovered.
Computer forensic science protocols should be written in a hierarchical manner so that overarching principles remain constant, but examination techniques can adapt quickly to the computer system to be examined. This approach to computer forensic protocols may differ from those developed for many traditional forensic disciplines, but it is necessary to accommodate a unique forensic examination.
American Society of Crime Laboratory Directors/Laboratory Accreditation Board (ASCLD/LAB). ASCLD/LAB Manual. American Society of Crime Laboratory Directors/Laboratory Accreditation Board, Garner, North Carolina, 1994, pp. 29–30.
CommerceNet Research Council. 2000 Industry Statistics. Available at http://www.commerce.net/research/stats/wwstats.html
Fischer, L. M. I.B.M. plans to announce leap in disk-drive capacity, New York Times (December 30, 1997), p. C-2.
Noblett, M. G. Report of the Federal Bureau of Investigation on development of forensic tools and examinations for data recovery from computer evidence. In: Proceedings of the 11th INTERPOL Forensic Science Symposium, Lyon, France. The Forensic Sciences Foundation Press, Boulder, Colorado, 1995.
Pollitt, M. The Federal Bureau of Investigation report on computer evidence and forensics. In: Proceedings of the 12th INTERPOL Forensic Science Symposium, Lyon, France. The Forensic Sciences Foundation Press, Boulder, Colorado, 1998.
Pollitt, M. Computer Evidence Examinations at the FBI. Unpublished presentation at the 2nd International Law Enforcement Conference on Computer Evidence, Baltimore, Maryland, April 10, 1995.