Scroll to turn the page
Cybersecurity · AP Cybersecurity
risk, controls and physical security
An adversary who can reach a device can bypass many of the technical controls that protect it. Securing the physical space is a first layer of defense.
The figure is an office (hypothetical). An employee badges through the door from the lobby; an adversary walks in close behind before it shuts. Nobody is watching the hallway, and the server room door has no lock. In 16.2 seconds the adversary is standing at the servers, without guessing a password or getting past a firewall.
To see the weaknesses in a space, think like an adversary: what is valuable here, who would want it, and which weakness would let them reach it?
This chapter is Unit 2 of AP Cybersecurity, Securing Spaces. It first sets out the ideas the whole course uses (2.1): who attacks and how, how to assess and manage risk, and how to choose security controls. Then it applies them to a building: physical attacks and vulnerabilities (2.2), protecting a space (2.3) and detecting a physical breach (2.4). The exam is weighted by skill, not by unit: analyzing risk, mitigating risk and detecting attacks are each 25–40% of the multiple-choice section.
pressure, applied on purpose
Social engineers use psychological tactics to manipulate targets into taking a desired action: revealing a password, holding a door, plugging in a drive.
The message in the figure (hypothetical) uses seven tactics, lit one at a time:
The tactics work together. Urgency and intimidation leave no time to check; authority and familiarity make checking feel rude.
who attacks, and why
Knowing who is likely to attack, and why, tells a defender what they will go after and how hard they will try.
Script kiddies are low-skilled adversaries who use tools developed by others without understanding how the tools work. They are often motivated by greed or a desire for recognition.
Hacktivists are motivated by social, political or personal causes. They compromise computers and networks to support their cause or stop a perceived harm, believing their goals justify their illegal methods.
Insiders are unique threats because they already have legitimate credentials and access to systems and data: in the figure the insider stands inside the wall. They can be recruited by malicious third parties and can be motivated by greed or revenge.
Cyberterrorists are motivated by politics or beliefs and seek to disrupt entire communities, regions or nations, for example by attacking a power grid, a water treatment plant or other civil infrastructure. They act independently or for governments or criminal organizations.
Transnational criminal organizations seek financial gain, mainly by deploying ransomware and by stealing corporate intellectual property (IP) to sell in illegal markets.
from looking to leaving no trace
Cyberattacks aim to disrupt, harm, steal or destroy devices, networks or data. Adversaries work in phases, though not every attack uses all of them.
| phase | what the adversary does |
|---|---|
| Reconnaissance | gathers information, often OSINT: freely available |
| Initial access | a foothold on one computer: a phish or a weak password |
| Persistence | a remote access trojan calls home over C2 |
| Lateral movement | on to accounts and computers with more privilege |
| Taking action | collects the target data and exfiltrates it |
| Evading detection | edits the logs and erases what it planted |
In reconnaissance the adversary gathers as much information as possible about the target, often open source intelligence (OSINT): staff pages, social media posts, job ads. For initial access they establish a foothold on one computer, often through social engineering or compromised or weak credentials.
To keep that access without having to regain it, they establish persistence, often with malware such as a remote access trojan (RAT) or rootkit that uses a command and control (C2) protocol to receive commands and send back output.
In lateral movement they escalate privileges by reaching computers and accounts with elevated permissions. Taking action, they collect the targeted data and exfiltrate it, disrupt services or destroy data. Finally, many try to evade detection by removing or editing log files and erasing files they planted.
what makes a risk
Risk occurs when a threat can exploit a vulnerability to compromise an asset.
An asset is anything valuable: financial resources, intellectual property, data, digital infrastructure, physical property and reputation. Here it is a server holding customer data.
A threat is what could cause the harm: a human adversary, or a natural disaster. A vulnerability is a weakness or flaw that could allow an asset to be compromised: here, a door with no lock.
In the first half of the figure all three line up: the threat walks through the unlocked door and reaches the server. In the second half a lock removes the vulnerability, and the same threat stops at the door. The asset is as valuable as before, and the adversary is as determined, but this way in is closed.
So risk assessment starts by listing assets, the threats to each, and the specific vulnerabilities a threat could use against them.
the two factors of risk assessment
Risk assessment considers two factors: the likelihood of an attack against a specific vulnerability, and the severity of the projected damage if it happens.
Likelihood depends on the value of the target (adversaries go after what they perceive as valuable), the skill required (well-documented exploits need less skill, so more adversaries can carry them out) and the motivation and capabilities of likely adversaries (skilled, motivated ones can perform more complex exploits).
Severity is often measured by financial cost, which can include reputational and operational impacts.
A qualitative assessment rates each risk in words. The figure places 5 risks to a veterinary clinic (hypothetical) on a grid of likelihood against severity; each cell’s rating combines the two, from low to severe:
| risk | likelihood | severity | rating |
|---|---|---|---|
| 1. ransomware on the records server | likely | high | severe |
| 2. front-desk laptop stolen | likely | moderate | high |
| 3. flood in the basement server closet | unlikely | high | medium |
| 4. website defaced by a hacktivist | possible | moderate | medium |
| 5. shoulder surfing at the front desk | possible | low | low |
Motive raises likelihood even without profit: a hacktivist who opposes the clinic’s policy gains nothing from defacing its website, yet it is a high-value target to them.
qualitative or quantitative
The result of a risk assessment can be qualitative (low, medium, high, severe; or “likely, high impact”) or quantitative: a number on a scale such as 1–10, or a quantifiable impact such as a dollar loss a year.
The figure gives the same 5 risks a quantitative rating, an estimated loss a year (hypothetical), as the height of each column. Numbers make risks easy to compare and rank:
| risk | qualitative | quantitative |
|---|---|---|
| 1. ransomware on the records server | severe | $120,000 a year |
| 2. front-desk laptop stolen | high | $30,000 a year |
| 3. flood in the basement server closet | medium | $20,000 a year |
| 4. website defaced by a hacktivist | medium | $12,000 a year |
| 5. shoulder surfing at the front desk | low | $2,000 a year |
Risk assessment documentation should include:
avoid, transfer, mitigate, accept
Once a risk has been identified and assessed, an organization has four options: avoid, transfer, mitigate or accept it.
The figure follows one risk (hypothetical): staff laptops holding customer records, an estimated $80,000 a year of loss from theft.
Avoidance stops the activity that generates the risk. The staff need their laptops to work, and if an activity is a critical part of the organization’s mission, avoidance is not possible.
Transference places the burden on another entity, such as an insurance company, a government or consumers. An insurance policy takes $30,000 of it.
Mitigation implements security controls to reduce the likelihood or impact of the risk. Encryption and cable locks remove $40,000.
What remains, $80,000 − $30,000 − $40,000 = $10,000, is the residual risk: the level of risk the organization is willing to accept. Acceptance acknowledges that absolute security is unattainable.
To conserve money and staff time, organizations favor controls that are cost effective and easy to install and maintain. A cost-effective control costs less to install and maintain than the expected loss from an attack: the controls here cost $6,000 a year and remove $40,000 of expected loss.
what a control protects
Every security control addresses at least one of three principles: confidentiality, integrity and availability.
| principle | ensures that | without it |
|---|---|---|
| Confidentiality | only authorized people, systems or processes can access the data | data theft or destruction |
| Integrity | the data are accurate and trustworthy | data manipulation |
| Availability | data and services are there for authorized users when needed | unexpected downtime |
The figure takes a customer record and breaks each principle in turn. Confidentiality fails when a copy leaves for someone not authorized to see it. Integrity fails when a value is changed: the balance reads $9,200.00 where it was $1,200.00, and the record can no longer be trusted. Availability fails when the record is there but authorized users can’t reach it.
Naming the principle at stake helps choose the control: encryption protects confidentiality, a hash check detects a loss of integrity, and a backup power supply protects availability.
what a control is, and what it does
Security controls can be classified by type, what they are, and by function, what they do.
By type: physical controls provide security in the physical space (locks, fences, cameras, bollards, security guards); technical controls provide security in the digital space (firewalls, anti-malware software, encryption); managerial controls are the rules, guidelines, policies and procedures that specify what security should be in place (password policies, regular access reviews, incident response plans, IRPs).
By function: preventative controls address potential vulnerabilities to stop an adversary from attacking (locks, encryption); detective controls help identify attacks when they occur (intrusion detection systems, IDSs; cameras; security information and event management, SIEM, systems); corrective controls fix problems and help restore systems to an operational state (vulnerability patching, repairing a broken card reader, intrusion prevention systems, IPSs).
Every control has both a type and a function. A camera is physical and detective; encryption is technical and preventative. The figure places one example in each cell.
layers, so one failure is not the end
A defense-in-depth strategy, or layered defense, uses multiple types of security controls to protect sensitive data and systems.
The layers can include human, physical, network, device, application and data. In the figure the data sit at the center and each ring is a layer.
The adversary gets through the first 3: the human layer fails when an employee holds the door; the physical layer is passed once they are inside the building; the network layer is passed when the adversary plugs into an open network jack. The device layer (locked workstation, host firewall) stops them, and the application and data layers were never touched.
Layering is necessary for two reasons. Different threats each need the control best suited to mitigate them: a guard does nothing against a phish, and a spam filter nothing against a tailgater. And when one control is bypassed, another may still prevent access or limit the damage: the strategy is resilient.
through a door that someone else opened
Adversaries often use social engineering in a physical attack. Two attacks get them through a door an authorized person has opened.
Tailgating is gaining unauthorized access to a restricted area by following close behind an authorized individual without that individual’s awareness or knowledge. In the first half of the figure the employee badges in and never looks back.
Piggybacking uses social engineering to manipulate an authorized individual into granting access. In the second half the adversary carries a large box, and the employee holds the door for them.
Common piggybacking tactics: carrying something large so an authorized person holds the door open, pretending to be an authorized person who has forgotten their access token, or pretending to be a maintenance worker who needs to get into an area for an inspection or repair.
The difference is the authorized person’s knowledge: unaware is tailgating; persuaded is piggybacking.
stealing what is left in view
Three more physical attacks collect what a target leaves exposed.
Shoulder surfing: an adversary watches as a user accesses sensitive information so they can use it later. Sometimes they use a camera to record the target for later analysis. In the figure the adversary reads the PIN typed at a desk.
Dumpster diving: an adversary goes through a target’s physical trash for information that could help reach their goal: a password on a sticky note, an org chart, a vendor’s invoice for a pretext.
Card cloning: an adversary makes a copy of an authorized user’s access card, so they can reach every resource that user is authorized to access. The copy carries the same card number, 0x3F2A91, so the door cannot tell the copy from the original.
loss, damage, disruption, destruction
Threats include human adversaries seeking to cause harm or disruption, and natural disasters. Vulnerabilities are weaknesses or flaws that could allow an asset to be compromised.
Common compromises: unauthorized access to sensitive data or restricted physical spaces; disruption of services; theft or destruction of digital or physical resources; and unauthorized modification of data.
Ports. With physical access to a device and its ports, an adversary can plug in a keylogger or an external drive containing malware, which could collect data from a user or even give them control of the device. They can also physically destroy the device, making it, its data and its services unavailable.
Power. When adversaries disrupt power to a device, the device and its services become unavailable. They may damage fuses or breakers in an electrical box, unplug or cut wiring, or damage substations and transformers.
Disasters. Natural disasters can damage or destroy computers and data and disrupt the digital services computers provide.
And an adversary who reaches an area with sensitive information can simply steal or copy it.
high, moderate and low
Physical access to devices can let adversaries bypass many technical controls and layers of security. So physical vulnerabilities are assessed and documented like any other risk.
High risks arise when sensitive information or systems are exposed in physical spaces without sufficiently restricted and controlled access. In the office: customer data on servers in a room with no lock, off an unmonitored hallway.
Moderate risks arise when a noncritical or nonsensitive part of an organization is left unprotected in a way that could give an adversary a foothold for initial access to other resources. In the office: the receptionist’s computer: outside the badge door, on the internal Wi-Fi, USB ports exposed.
Low risks arise when a vulnerable asset is of low value and the vulnerability is unlikely to be exploited. In the office: laptops with no sensitive data, left on desks at lunch, inside the badge door, no cable locks.
The receptionist’s computer holds nothing important, but it is a way in: what makes it moderate is what it connects to.
training and a workstation policy
Many physical protections are managerial: rules people follow. Two matter most.
Security awareness training teaches employees how they contribute to security: detecting social engineering attempts such as phishing, not badging other people into restricted areas, and preventing device theft.
A workstation security policy sets out the measures that protect a physical workplace. It may have tiers, based on the type of data handled at a workstation. In the figure the desk breaks every rule, then the policy is applied one rule at a time:
deter, prevent, and admit one at a time
To choose a control, a defender considers how an adversary could use a vulnerability to attack, and how to prevent, detect or correct the attack.
Fencing, gates and bollards around a building deter adversaries from trying to physically access it. Locks on doors, server cabinets and computers prevent devices being accessed or stolen. Card readers record which badges are used at which entries and when, and deny unauthorized badges.
An access control vestibule (two doors, only one open at a time) and turnstiles prevent an authorized person from intentionally or accidentally admitting an unauthorized one. In the figure two people step in on one badge; the inner door will not open with two inside, and the adversary has to leave.
Organizations can disable USB ports so external drives can’t load malware. An uninterruptible power supply (UPS) gives a device backup power in an outage; generators power a building or a set of critical devices.
Mitigations are prioritized by the severity of the risks and the cost of the mitigations (hypothetical figures):
| risk | severity | mitigation | cost |
|---|---|---|---|
| server room has no lock | high | badge lock on the server room door | $1,800 |
| tailgating at the badge door | moderate | an access control vestibule | $24,000 |
| tailgating at the badge door | moderate | awareness training and a door-open alarm | $2,500 |
| receptionist PC USB ports open | moderate | disable the USB ports | none |
| laptops loose on desks | low | cable locks | $900 |
cameras, sensors, guards and people
Detective controls work only where they are placed well.
Cameras capture a visual record of an adversary’s activity; the feed should be recorded and monitored, and recordings help in after-incident investigations. Place them for coverage and angle, out of an adversary’s reach, where they would capture something useful. Points of ingress and egress are often watched: here the entrance, the back exit and the server room.
Motion sensors alert security to movement. Place them where traffic is unexpected, like a server room, or where sensitive materials are stored and few people have access. In a busy area they raise so many false alarms that a real one is less likely to be taken seriously. They work best paired with cameras: when the server room sensor fires, the server room camera confirms the breach.
Security guards monitor an area and respond. Stationary guards give constant protection to an area, entrance or high-value item, and are most effective where traffic funnels (entry gates, main entrances and lobbies, entrances to more secure areas). Patrolling guards are harder to plan around, create time pressure, and suit perimeters and exterior areas.
Employees who work in a space are often the first to notice someone who doesn’t belong, and can alert security.
a door open too long
When employees must badge to unlock a restricted door, a sensor can record how long the door stays open. A door open longer than normal is a sign of piggybacking or tailgating.
The server room door’s log for one day (hypothetical):
08:02:11 door SRV-1 badge 1047 open 3.1 s 08:15:40 door SRV-1 badge 1122 open 2.8 s 09:01:05 door SRV-1 badge 1047 open 3.4 s 10:22:37 door SRV-1 badge 1309 open 11.6 s 11:48:02 door SRV-1 badge 1122 open 3.0 s 12:30:19 door SRV-1 badge 1185 open 2.6 s 13:05:44 door SRV-1 badge 1309 open 9.8 s 14:12:56 door SRV-1 badge 1047 open 3.3 s
A typical opening lasts about 3.2 s (the median). Two openings, at 10:22:37 and 13:05:44, lasted 11.6 s and 9.8 s, more than the 6 s limit set for review, and both used badge 1309: one badge, but time for more than one person.
Then defenders turn to the cameras. Live and recorded footage tracks an adversary’s path and actions once a breach is detected, and cameras paired with facial recognition software can raise an alert when an unauthorized individual enters a controlled area.