Physical Protection: Barriers, Threat Models, and Lock Mechanics

Physical protection is a fundamental and critically important layer of security for any distributed information system. Computer security often operates with abstractions (encryption, protocols, access control), yet all these mechanisms function on physical media – servers, cable routes, hardware modules, and user terminals. If an attacker gains unrestricted physical access to equipment, most logical security barriers can be bypassed or completely destroyed.

Introduction to Physical Security and the Concept of Defense in Depth

The main task of physical protection is to limit or prevent an attacker’s unauthorised physical contact with critical system components. In engineering practice, security is never built around a single super‑reliable perimeter. Instead, the concept of Defense in Depth is applied.

Its essence is the creation of successive concentric security zones:

1. Outer perimeter: Site fencing, gates, anti‑ram barriers, external video surveillance.
2. Inner perimeter: Building walls, reinforced windows, access control points at entrances.
3. Restricted access zone: Floors, corridors, or wings of the building protected by gateways and access control systems.
4. Hardware zone: Server rooms, protected racks and cabinets with tamper sensors.
5. Device level: The server enclosures themselves, cryptographic processors (TPM/HSM) with lid‑removal sensors and key‑erasure circuits triggered upon opening.

The main engineering principle of layering: the time required for an attacker to overcome all physical barriers must be guaranteed to be greater than the time required for the response force (guards) to detect the attack and arrive on site.

Threat Model in the Physical Environment

Designing protection systems begins with formalising a threat model. Physical threats differ substantially from network threats because they are constrained by the laws of mechanics, thermodynamics, and the presence of a human being in space. Classification of intruders by level of motivation, resources, and skill:

• Vandals and Opportunists: Act spontaneously, use simple available tools (stones, crowbars). Their goal is petty theft or destructive acts. Protection against them is trivial and amounts to basic sturdy construction.
• Skilled Burglars / Insiders: Have a clear target (e.g., theft of a specific server containing confidential data). They possess specialised tools (diamond cutters, hydraulic tools, lock picks), know the building topology, and may have insider information about guard schedules or vulnerabilities in the access control system.
• State‑Sponsored Actors / Intelligence Agencies: Possess practically unlimited resources. They are capable of using covert entry techniques (electronic hacking, producing duplicate keys from high‑resolution photographs), employing sophisticated social engineering, infiltrating agents into maintenance staff, or intercepting equipment during the supply chain (Supply Chain Attacks).

Economics of Deterrence

Physical security is rarely absolute. Its effectiveness is assessed economically: the cost for the attacker to overcome a barrier (including tool costs, time, risk of being caught, and criminal liability) must significantly exceed the potential gain from theft or data compromise.

Walls and Barriers: Mechanics of Material Resistance

A physical barrier is designed to create a time delay. In international practice, building structures and obstacles are classified according to burglary resistance classes (e.g., EN 1627–1630 or UL standards).

When calculating barrier strength, engineers take into account the following parameters:

• Thickness and material composition: Ordinary brick can be breached with hand tools in minutes. Reinforced concrete requires rotary hammers and plasma cutters. For server rooms of the highest reliability category, walls are reinforced with steel sheets or specialised composite panels.
• Joints and weak points: A strong wall by itself is useless if the junctions between wall and ceiling, ventilation shafts, or cable entries are not protected. Attackers often enter premises through false floors or false ceilings of adjacent common‑access rooms.
• Doors and window openings: Door leaves must match the strength class of the walls. Door frames are rigidly anchored into the walls to prevent jacking out. Windows in server areas are either completely eliminated or protected with armoured glazing and steel grilles welded to the building’s reinforcement.

Mechanical Locks: Anatomy and Vulnerabilities of Master Key Systems

A lock is a mechanical or electromechanical device that blocks the movement of the bolt relative to the door frame. Understanding the operating principles and attack vectors on mechanical locks is critically important, because they still remain the last or primary line of defence for doors and server racks.

Anatomy of a Pin‑Tumbler Lock

The most common type of mechanical lock is the pin‑tumbler lock. It consists of:

1. Plug (actuator): The rotating core into which the key is inserted.
2. Stator (cylinder): The fixed outer part of the lock.
3. Shear Line: The boundary between the radial surfaces of the plug and the stator. If pins cross this line, the plug is blocked and cannot rotate.
4. Pins: Divided into pairs. The key pins contact the key and have different lengths. The driver pins are spring‑loaded and sit above them.

When the correct key is inserted, its profile lifts each pair of pins to exactly the height at which the joint between each key pin and driver pin aligns with the shear line. The plug can then rotate freely, moving the cam that opens the bolt.

Technical Picking Methods and Mechanical Vulnerabilities

1. Classic Lockpicking (Single‑Pin Picking): Due to inevitable microscopic manufacturing defects (tolerances, hole misalignment), the pins in a lock are never in a perfect straight line. The attacker inserts a tension wrench and applies a slight rotational force to the plug. One pin binds (jams) more than the others. Using a pick, the locksmith carefully lifts that particular pin until they feel a slight click – the driver pin has crossed the shear line and become caught on the ledge due to the plug’s misalignment. The process is repeated for all pins one by one.
   * Countermeasures: Use of pins with complex shapes (mushroom, spool, serrated pins) that create false sets and bind the plug during picking attempts.
2. Bumping (Kinetic Attack): A highly effective kinetic attack. The attacker makes a so‑called bump key – a key with the deepest possible cuts for all pin positions. The key is inserted into the lock almost fully, a light rotational force is applied to the plug, and then a sharp blow is struck to the key head (e.g., with a special plastic hammer). The impact impulse is transmitted from the key to the key pins and from them to the driver pins. According to the law of conservation of momentum (Newton’s cradle effect), the key pins stay in place while the driver pins fly upward, compressing their springs. For a fraction of a second, a gap appears between the pins exactly at the shear line. At that moment, the torque rotates the plug and the lock opens. The attack takes seconds and leaves almost no external traces.
   * Countermeasures: Installation of magnetic elements inside the cylinder, use of telescopic pins (pin‑in‑pin systems like Mul‑T‑Lock), addition of interactive moving elements on the key (floating balls, floating pins), use of side serpentine tracks (disc‑detainer locks such as Abloy Exec/Protec, which are completely immune to bumping).
3. Impressioning: A method of creating a duplicate key from an existing lock without disassembly. A blank key made of soft metal (e.g., brass) is inserted into the lock, clamped in a vice, and rocked up and down. Where the pins bind the plug, they leave microscopic marks on the blank. The attacker removes the blank, files down the metal at the marks with a needle file, and repeats the process. When all the cuts have been filed to the correct depth, marks no longer appear, and the lock opens.

Vulnerabilities of Master Key Systems

Organisations often deploy hierarchical mechanical access systems – master key systems. They allow an employee to open only their own office with their key, while a manager or security officer can open all doors in the building with a single master key.
To implement this logic, a third, intermediate pin – a master wafer – is placed between the key pin and the driver pin in each pin chamber. The presence of this wafer means that each pin chamber now has two alternative shear points at which the plug can turn.
Let us describe the fundamental vulnerability of such an architecture from an information‑theory perspective:
If a lock uses n pin chambers, and each chamber contains a master wafer, then instead of one correct code, the lock begins to accept 2ⁿ different key combinations.

This gives rise to two serious threats:

1. Cryptanalysis / Combination Attack: Having just one legitimate low‑level key (e.g., to one’s own room), an attacker can analyse its geometry, calculate the cut steps, and by controllably altering the heights of individual pin positions (making a series of test keys), quickly reconstruct the geometry of the Master key. The master system sharply reduces the entropy of the lock.
2. Cross‑Keying (Accidental Matches): Due to wear and wide tolerances in cheap master systems, a key from one office may unexpectedly open an office on another floor, because the combination happens to match one of the many permitted intermediate states of the cylinder.

Electronic Locks, Alarm Systems, and Methods of Bypassing Them

The transition from purely mechanical protection to electronic and combined systems automated access control, but gave rise to new attack vectors lying at the intersection of physics, circuit design, and programming.

Electronic and Biometric Locks: Access Control System Architecture and Vulnerabilities

Modern electronic locks and Access Control Systems (ACS) replace a mechanical secret with a digital identifier. A typical ACS architecture includes:

1. Token: A contactless RFID/NFC card, magnetic card, or biometric feature.
2. Reader: A device that reads the code from the token and transmits it to the controller.
3. Controller: A microprocessor module that makes an access decision based on a rights database.
4. Actuator: An electromagnetic lock (holds the door by magnetic force) or an electromechanical latch/solenoid.

Technical Vulnerabilities and Attack Vectors

• Attacks on the Wiegand Interface: Historically, most readers connect to the controller via the Wiegand protocol (developed back in the 1970s). This is a simple three‑wire interface (Data 0, Data 1, Ground) that lacks encryption, authentication, or integrity checks.
• Attack mechanics: By removing the external reader housing, which is located on the unprotected (outside) side of the door, an attacker gains direct access to the Wiegand wires. By connecting a hardware sniffer (e.g., based on an ATmega microcontroller or a specialised tool like ESPKey), they can intercept a legitimate employee’s card ID in plain text and then replay it to open the door (Replay Attack).
• Vulnerabilities of Contactless Cards (RFID/NFC): Low‑frequency cards (125 kHz, such as EM‑Marine or HID Prox) have no cryptographic protection at all. They continuously broadcast their fixed UID over the air when in the reader’s field. Duplicating such a card takes seconds using compact devices (Proxmark3 or Flipper Zero) from a distance of up to several tens of centimetres from the victim in a public place.
  High‑frequency cards (13.56 MHz, e.g., older generations of MIFARE Classic) use proprietary encryption algorithms (CRYPTO1) that have been fully reverse‑engineered. Mathematical vulnerabilities (Nested and Hardnested attacks) allow secret sector keys to be recovered in minutes with only partial access to the reader or the card itself.
• Attacks on Actuators (Solenoids and Magnets): Electromagnetic locks hold the door by magnetic force (from 200 to 500+ kg). However, if a thin layer of transparent tape or non‑conductive tape is discreetly placed between the magnet and the mating metal plate, the holding force drops by 70‑80%. Externally, the lock will appear locked (the indicator lights green), but the door will open with a strong, sharp jerk.
  Electromechanical solenoid locks (with a spring‑loaded metal core) are vulnerable to kinetic and magnetic attacks. If the lock body is located close to the door surface, a powerful neodymium magnet applied externally to a specific point on the door leaf can overcome the internal spring force, retract the solenoid, and unlock the bolt without any electrical signal being applied.

Alarm Systems: The Physics of Sensors

Alarm systems are designed to detect the very fact of an intruder breaching the perimeter. Their reliability depends on the physical principles underlying the sensors. Let us examine the main types of sensors and the physical limitations that make their bypass possible.

Classification of Sensors by Physical Principle

1. Magnetic Reed Switches: Installed on windows and doors. They consist of a sealed glass capsule with ferromagnetic contacts inside (on the frame) and a permanent magnet (on the moving sash). When the door is closed, the magnet holds the contacts closed. When opened, the field disappears, the circuit opens, triggering an alarm.
2. Passive Infrared (PIR) Sensors: Detect the thermal radiation of objects in the infrared range (wavelength about 10 µm, corresponding to human body temperature). A Fresnel lens is placed in front of the pyroelectric element. It divides the field of view into alternating sectors (beams) and blind zones. When a person moves across these sectors, the sensor detects a rapid change in heat flow (differential signal) and generates an alarm.
3. Microwave (Radar) and Ultrasonic Sensors: Active volumetric sensors. They emit high‑frequency radio waves (usually 2.4 or 10.5 GHz) or ultrasound into the room, then receive the reflected signal. They operate on the Doppler effect: if there is a moving object in the room, the frequency of the reflected wave changes proportionally to the object’s speed.
4. Vibration‑Seismic and Acoustic Sensors: Acoustic sensors are tuned to a specific frequency spectrum: first they detect a low‑frequency pressure wave (impact on glass), then a high‑frequency ring (glass shattering). Vibration‑seismic sensors are mounted directly on walls or metal grilles to detect attempts to cut or destroy them with a rotary hammer.

Sensor Defeat Methods

Experienced intruders exploit the physical limitations of sensors to prevent the generation of a differential signal or to keep environmental parameters within the normal range.

• Bypassing Reed Switches (Magnetic Field Substitution): If the attacker knows the exact location of the reed switch on the door (which is easy to determine from standard installation practices), they can carefully apply a strong external magnet to the outside of the door directly opposite the hidden reed switch. The door can then be opened: the internal legitimate magnet moves away with the sash, but the reed contacts remain closed under the influence of the attacker’s external magnet. The alarm loop is not broken.
• Bypassing PIR Sensors (Thermal Masking): A PIR sensor reacts not to heat itself, but to the temperature difference between the human body and the background, multiplied by the speed of movement between the lens sectors.
• Bypass technique: If the air temperature in the room approaches 35‑37°C (e.g., on a hot summer day in a server room when the air conditioning fails), the effectiveness of PIR sensors drops to almost zero because the thermal contrast disappears. Attackers also use shielding suits – thick cloaks, thermal insulation blankets, or large sheets of architectural foam held in front of them. If they move extremely slowly (less than a few centimetres per second) directly towards the sensor (rather than across the Fresnel lens beams), the sensor will interpret the temperature change as a natural slow background drift and ignore it.
• Bypassing Microwave Sensors: Microwave‑band radio waves can penetrate thin walls, drywall, and glass. However, they are completely reflected by metals and blocked by materials with high water content. An attacker can block the active sensor’s field of view by discreetly placing an obstacle in front of it (e.g., during a legitimate daytime visit under the guise of furniture maintenance or cleaning, positioning a cabinet or large box), creating a permanent blind spot for night‑time intrusion.

Case Study: How Not to Protect a Painting

Let us cite a textbook example of a physical security design failure from real museum robbery practice. A valuable painting was protected by a complex combination: a vibration sensor behind the canvas, a PIR sensor in front, and an infrared barrier.

• Design error: All sensors were wired to a single controller, whose control lines ran behind a false wall.
• Attack execution: The attackers gained access to the building through a ventilation shaft that was outside the PIR sensor’s field of view. They did not approach the painting directly. Instead, they drilled a small hole in the wall from an adjacent general‑access utility room and poured expanding polyurethane foam into the alarm junction box, physically jamming the mechanical relays that held the loops in a closed state. Once the relays stuck in their positions, the sensors could trigger alarms as much as they wanted, but the signal physically could not reach the security monitoring station.

Feature Interactions and Attacks on Communication Lines

A serious vulnerability of complex security systems is unpredictable interaction between different features, especially at the junction of fire safety systems, ACS, and intrusion alarms.

Conflict of Interest: Life Safety vs. Asset Protection

Under international building codes (e.g., NFPA), fire safety requirements always take absolute priority over burglary protection requirements. In the event of a fire alarm (or loss of building power), all electromagnetic and electromechanical ACS locks on evacuation routes are required to automatically de‑energise and switch to the open state (Fail‑Safe), so that people are not trapped inside a burning building.

• Attack vector: Attackers do not need to break card cryptography or pick server‑room door locks. It is enough to simulate a fire – for example, activate a smoke detector in the corridor using a special aerosol, or press an external emergency door release button (Break Glass unit) that is required by law at every emergency exit. The automation system instantly removes power from all locks in the building, giving the attacker free access to the servers.

Attacks on Data Transmission Lines (Line Sabotage)

Signals from sensors are transmitted to the control panel via wired loops. In simple systems, the logic is closed/open.
To protect against simple wire cutting (which would cause an open circuit and an alarm), engineers use End‑of‑Line (EOL) resistors. In this case, the panel measures the exact analogue resistance of the loop (e.g., 4.7 kΩ in the normal state). If the wire is cut, the resistance becomes infinite (alarm Break). If the wire is short‑circuited, the resistance drops to zero (alarm Short).

• Engineering bypass: A wired loop is vulnerable if an attacker can physically connect to it at an intermediate point (e.g., in a junction box in the corridor). By measuring the line’s resistance with a multimeter in normal mode, the attacker connects a variable resistor in parallel with the line, adjusts it to the equivalent resistance value (the same 4.7 kΩ), and then physically cuts the wire going to the real sensor. The central panel does not notice any change in the line current, recording an ideal normal loop state, while the attacker penetrates through the disabled sensor.
• Countermeasures: Transition from analogue loops to addressable digital buses (e.g., RS‑485 or CAN), where each sensor continuously exchanges cryptographic data with the controller using dynamic keys and message counters (to protect against Replay attacks).

Monitoring and Resource Accounting: Smart Metering, Taximeters, and Fiscal Systems

In information systems, monitoring and metering represent critical functions for ensuring trust. While classic security systems (alarms) protect the confidentiality of space or the integrity of the perimeter based on the availability of an alarm signal, commercial metering systems face a far more aggressive physical environment.
The main feature of the threat model here is that the end legitimate user of the system is simultaneously a potential violator. A consumer of electricity is motivated to stop the meter, a truck driver – to hide speeding or overtime, a taxi driver – to inflate the mileage reading. Let us analyse distributed metering systems as economic and cryptographic mechanisms operating under conditions of complete distrust of the hardware node.

Prepayment Meters and the Economics of Tokens

Historically, utility metering (gas, electricity, water) was built on a post‑payment scheme, where a company inspector read the mechanical accumulators once a month, after which the customer was billed. But for unreliable categories of consumers, persons without a fixed address, and those with no credit rating (e.g., in developing countries or poor districts of the UK), the only option is prepayment.
Evolution of prepayment systems:

Coins → Magnetic cards → Smart cards / Chip keys → Digital tokens (Magic Numbers)

Using physical coins is inefficient due to huge operational costs for collection and the risks of bill‑acceptor hacking. Modern smart metering is based on transferring value through digital tokens.

The Superposing Card Attack – A Physical Attack

An example of a trivial but effective hardware bypass of card‑top‑up systems, using payment terminals in corporate canteens. When topping up a card, the machine read the old value, accepted a coin, and wrote a new value.

• Attack mechanics: The attacker cut the corners off a legitimate card with a balance (e.g., £49), placed it on top of an empty card, and inserted the stack into the reader. The mechanical positioning levers inside the machine read the geometry of the lower (full‑size) card, but the reader’s electrical contacts touched the chip of the upper card. After inserting £1, the machine wrote the new value (£50) to the lower card. In the end, the attacker had two cards with a total balance of £99.
• Engineering conclusion: No cryptographic encryption of data on the chip can prevent this attack, because both read/write operations were absolutely legitimate from a cryptographic standpoint. The vulnerability lay at the interface between the mechanics and the logic of the reader’s finite state machine.

Architecture and Cryptography of STS Tokens (Standard Transfer Specification)

To eliminate expensive physical media, South Africa and several Latin American countries introduced a system of magic numbers, later standardised as STS. A customer buys electricity at a store or via SMS and receives a 20‑digit numeric code, which they enter manually through the meter’s keypad.
Token structure and STS cryptographic protocol:

1. Data Payload: Instruction class (top‑up, service command), amount of resource purchased (in kWh or currency).
2. Unique Meter ID: Guarantees that the token can only be activated on the target device.
3. Timestamp / Token Counter (Token ID): A monotonically increasing value that prevents Replay Attacks. If the meter sees a token with a time value less than or equal to one already stored in secure non‑volatile memory (EEPROM), the token is rejected.
4. Cryptographic Signature / Ciphertext: All data is packaged and encrypted using a block cipher (initially modified DES, later AES) under the device‑unique key K_ID.

Key Diversification

Storing one master key in all meters is impossible: compromising one device would destroy the security of the entire network. Therefore, a hierarchical key generation scheme is used. On the regional server (or in the secure module of the vending terminal), a regional master key K_V is stored. The unique key of a specific meter K_ID is computed as the encryption of the meter’s serial number with the master key:

K_ID = encrypt(serial_number, K_V)

When generating a token, the terminal takes the meter’s serial number, recovers its unique key K_ID, encrypts the top‑up data with it, and translates the binary code into a 20‑digit decimal sequence using a special error‑correcting encoding (to minimise user input typos).

Utility Smart Metering: Scaling Problems and Fraud Vectors

When commercial metering systems scale up to the size of a country, systemic vulnerabilities arise related to the absence of a back‑channel and market deregulation.

The Problem of No Back‑Channel

Simple meters with manual code entry (STS) or one‑way magnetic cards cannot transmit data back to the supplier. This gives rise to specific fraud vectors:

• Bypassing the meter (physical shunt): Physical shunting of current circuits. The attacker installs a powerful copper cable in parallel with the meter’s terminals. Current flows along the path of least resistance, and the meter’s current transformer or shunt registers only a small fraction of the actual consumption.
• Tariff manipulation: If the meter must support multi‑tariff metering (day/night), it requires accurate internal clocks. In the absence of network synchronisation, attackers use service engineering tokens to roll back the device’s internal clock, forcing it to count consumption 24/7 at the lowest night tariff.

Deregulation and Attacks on Distributed Vending Networks

In the UK, deregulation of the energy market led to a situation where one company (the Grid Network) owns the physical infrastructure, while dozens of competing retailers sell electricity to end users.
This complicated the key management architecture:

• A need arose for secure transfer of unique meter keys K_ID from the infrastructure owner’s database to terminals of third‑party commercial companies.
• If the key transfer protocol between retailers’ servers is not protected by end‑to‑end encryption, there is a threat of intercepting the key database. Extraction of master keys allows attackers to set up a shadow business generating illegal top‑up tokens (black vending).

Tachographs, Taximeters, and Speed Limiters: Combating Sabotage

Vehicle control is another critical area for metering systems. Here the main task is to record legally significant evidence of violations.

Mechanical Tachographs and Methods of Their Falsification

The classic analogue tachograph recorded vehicle speed and driver rest periods on paper wax discs (tachodiscs) using mechanical recorders. The speed sensor was installed on the gearbox and transmitted analogue electrical impulses.
Methods of bypassing mechanical tachographs were primitive but widespread:

1. Power sabotage: Brief removal of the instrument’s fuse during a night trip. The tachograph recorded no movement (rest) while the vehicle was actually travelling.
2. Magnet attack: A powerful magnet was placed on the gearbox impulse sensor, blocking the rotation of internal elements or distorting the induction signal. The counter received no movement impulses.
3. Mechanical interference: Pressing on the recorder’s stylus through technical gaps in the housing, or placing a rubber band under the disc, which resulted in drawing a perfect straight line simulating the regulatory speed of 80 km/h.

Digital Tachographs: Cryptography and Architectural Failures

To combat widespread fraud, the European Union initiated the Digital Tachograph Project. The system architecture includes a secure on‑board unit (VU – Vehicle Unit), driver and inspector smart cards, and an encrypted motion sensor.
Cryptographic basis of the digital tachograph:

• Each motion sensor on the gearbox is cryptographically paired with the VU using an authentication protocol based on elliptic curves or RSA.
• Each motion impulse is accompanied by a cryptographic tag (MAC – Message Authentication Code). Attempts to send fake impulses with a signal generator are rejected, because the attacker does not have the sensor’s session secret key.
• Driver activity data is digitally signed by the VU and stored in the protected memory of the driver’s smart card.

System‑Level Problems of Digital Tachographs

Despite strong cryptography, digital tachographs demonstrated classic failures in distributed security system design:

1. Signal duplication attack on the sensor: Drivers and illegal workshops discovered that if a powerful high‑frequency interference signal is applied to the motion sensor, the sensor’s protected microcontroller enters a hardware fail‑safe mode (Fail‑Silent). The VU records a sensor communication error but cannot prove the fact of motion. Legally, the driver can claim the device broke down on the road, and cannot be fined for overtime.
2. The collusion problem and the Market for Lemons: Transport companies are strongly interested in violating speed limits to speed up logistics. A closed market for specialised equipment to bypass the systems emerged. Equipment manufacturers (VU) unofficially compete with each other on the parameter of ease of covert sabotage for the end customer (the transport company), which in economics is known as information asymmetry and degradation of the security market.
3. Vulnerabilities of speed limiters: The speed limiter takes data from the tachograph and forcibly reduces fuel supply to the engine via the ECU (Electronic Control Unit) when 90 km/h is reached. Drivers learned to install a toggle switch in the data link between the VU and the ECU. When the line is broken, the ECU switches to a backup algorithm, allowing the vehicle to accelerate to its maximum design speed, although the tachograph itself logs a CAN bus error.

Taximeters as Fiscal Registrars

Historically, the first mass‑produced transport metering devices were taximeters. Their task is to convert distance travelled and idle time into monetary value according to an officially approved tariff.

Architecture and Data Sources

Early taximeters were purely mechanical and connected via a flexible cable to the gearbox output shaft or wheel hub. Modern electronic taximeters integrate into the vehicle’s on‑board network and receive data from two key sources:

1. Vehicle Speed Sensor (VSS): Generates pulses (usually via the Hall effect), the frequency of which is proportional to the rotation speed of the transmission output shaft.
2. CAN bus (Controller Area Network): The taximeter reads broadcast data packets on vehicle speed and engine RPM directly from the digital bus connecting the ECU units.

Classic Manipulation Vectors (Taxi Fraud)

Taxi drivers have invented many methods to bypass electronic metering systems in order to overcharge passengers (rolling the meter):

• Pulse generators (twisters): A hidden low‑frequency generator (e.g., based on an NE555 timer) is installed in the break of the signal wire from the speed sensor to the taximeter. While stuck in traffic or waiting for a passenger, the driver activates the generator, which simulates high‑speed driving, forcing the meter to increase the fare faster.
• Geometry manipulation (Wheel radius): The taximeter is calibrated to a strictly defined pulse index per kilometre (w‑index). This index is rigidly tied to the external diameter of the standard tyres. Installing tyres with a reduced profile or significantly lowering tyre pressure causes the wheel to make more revolutions to cover the same distance. The taximeter records a greater virtual mileage than the actual distance travelled.

The Transition to Digital Tachographs and High‑Tech Attacks

The large‑scale sabotage of analogue systems forced the European Union to completely overhaul the regulatory and engineering framework. The result was the international Digital Tachograph Project, designed to replace the easily vulnerable paper discs and analogue sensors with secure microprocessor nodes and cryptographic protocols.

Architecture and Cryptography of the Digital Tachograph

The digital tachograph is designed as a distributed hierarchical system whose security relies on a Public Key Infrastructure (PKI) and Common Criteria standards (target assurance level not lower than EAL4).
The system consists of three key elements:

1. Vehicle Unit (VU): A protected device built into the dashboard, containing fiscal non‑volatile memory and a cryptographic coprocessor.
2. Smart cards (Tachograph Cards): Driver cards (recording work regimes), workshop cards (calibration), inspector cards (control), and company cards.
3. Cryptographic motion sensor (e.g., KITAS family): An intelligent sensor installed on the gearbox.

Mutual Authentication and Pairing Protocol

When the KITAS motion sensor is first installed in a vehicle, a certified workshop performs a pairing procedure with the VU.
The information exchange process between the sensor and the VU is as follows:

• PKI‑based authentication: The sensor and VU exchange their digital certificates issued by the national Certification Authority (CA). They verify each other’s certificates using the state’s root public key, which is hard‑coded into their memory at the factory.
• Session key generation: Using the Diffie‑Hellman protocol (or similar asymmetric mechanisms), the sensor and VU generate a temporary symmetric session key K_session.
• Cryptographic protection of the data stream: Each physical impulse recording the rotation of the transmission shaft is transmitted not as a simple voltage change, but as an encrypted data packet or accompanied by a Message Authentication Code (MAC) computed with K_session.

Packet = {Sequence No, Pulse Count, Timestamp} MAC(K_session)

This completely eliminates the classic twister or simple frequency generator attack: the VU will instantly discard any impulses that are not signed with the correct current session key.

Systemic Vulnerabilities and High‑Tech Attacks on KITAS

The introduction of cryptography shifted the attack vector towards the physical layer and hardware failures, confirming the thesis: Attackers do not attack cryptography, they attack the environment in which it operates.

The FAIL‑SILENT Attack via Electromagnetic Suppression

Although the cryptographic core of the KITAS sensor cannot be broken, its analogue part (the physical Hall sensor reading the rotating magnetic gear) remains subject to the laws of physics.

• Attack mechanics: Drivers discovered that if a specially modified powerful high‑frequency electromagnetic field generator is placed on the housing of a modern KITAS sensor, the interference penetrates the sensor’s internal shielding. The sensor’s microcontroller, detecting critical data corruption on its internal buses, activates Fail‑Silent mode (safe silence) and stops sending packets.
• Result from the system’s perspective: The VU logs a Sensor communication fault event. The vehicle continues moving, but the VU receives no cryptographic motion data. The driver gains the ability to continue driving at night, and can show the inspector a technical device fault, for which the law does not allow licence revocation.

Attacks on Calibration Settings (w‑index)

To calculate speed correctly, the VU must know how many sensor impulses correspond to one kilometre of travel (the vehicle’s w‑index). This value is written to the VU by a workshop technician using a Workshop Card during periodic calibration on roller benches.

• Fraud vector via compromised workshops: Workshop cards are protected by PIN codes, but a black market of corrupt service centres emerged in the transport industry. The technician inserts a legitimate calibration card and forcibly inflates the w‑index in the VU’s memory. As a result, the tachograph believes the vehicle is travelling at 85 km/h, while its actual speed is 100 km/h. The speed limiter does not activate because it fully trusts the false calibration data from the VU.

The Resurrecting Duckling Protocol

Having analysed the problems of secure device binding in transport and embedded systems, Ross Anderson, together with Frank Stajano, developed a fundamental concept – the Resurrecting Duckling protocol. This protocol solves the problem of secure device pairing in conditions where devices have no permanent access to the global network (the internet) or a trusted third party (PKI).

Philosophy and Metaphor of the Protocol

The protocol is based on the biological phenomenon of imprinting: a newly hatched duckling considers the first moving object it sees (usually its mother) as its parent.
In the digital world:

• Duckling: A peripheral device (sensor, smart lock, tachograph sensor) that has no input interface of its own (keyboard/screen).
• Mother: The controlling controller (tachograph VU, owner’s smartphone, diagnostic computer).

Phases of the Device Lifecycle

1. Egg state (Pre‑birth): The device is manufactured at the factory, its memory is empty, it has no owner. It is ready to accept any first correct key.
2. Imprinting (Birth): The controller (Mother) connects to the device via a physical or secure short‑range channel (e.g., cable or NFC) and transmits a secret key K_M (Master key) to it. The duckling stores this key in its protected non‑volatile memory. From this moment on, the duckling obeys only that specific mother. All requests from other controllers are completely ignored.
3. Life and Obedience: During operation, the Mother can issue commands to the duckling or temporarily delegate control rights to other nodes (e.g., the Mother VU allows the duckling sensor to exchange data with a diagnostic station by creating a temporary session key).
4. Death and Resurrection: What happens if a vehicle is scrapped, the sensor needs to be moved to another vehicle, or the owner changes? The Mother sends the duckling a specially cryptographically signed Die command. Upon receiving this command, the duckling’s microcontroller initiates a guaranteed destruction (reset) of the stored key K_M in its memory. The device returns to the initial egg state (cleared and ready for new imprinting with a different Mother).

Application in Monitoring Systems

The Resurrecting Duckling protocol eliminates the need to deploy complex, expensive government PKI servers for every small industrial sensor. It guarantees that as long as the sensor is physically connected to a specific on‑board unit, their link is exclusive and protected against Man‑in‑the‑Middle (MITM) attacks throughout the entire operational lifecycle.

Postage Meters and Final Technical Synthesis

Postage meters represent one of the most interesting examples of commercial metering systems, because they effectively function as local printers of value‑equivalent stamps (postal currency) directly on the end‑user’s premises. This fundamentally changes the threat model compared to classical IT systems.

Postal Postage Meters: Evolution of Protection and Architecture

Historically, postage meters were developed to automate the sending of large volumes of mail by large companies. Instead of buying and manually sticking millions of physical postage stamps, organisations gained the right to use secure printing presses that applied a payment imprint (franking) directly onto the envelope. The postal authority would either be prepaid or invoiced based on the readings of the meter’s internal counters.

Evolution of Protection Mechanisms

Mechanical stamps (lead seals)
       │
       ▼
Electromechanical systems (fiscal registers)
       │
       ▼
IBI (Information‑Based Indicia) cryptographic stamps

In the mechanical era, protection relied exclusively on physical barriers: strong metal housings, high‑security tubular locks, and disposable lead seals. A postal inspector personally visited the company’s office, broke the seal, set the mechanical register to a certain amount (e.g., £1000) after receiving payment, closed the housing, and installed a new seal. As stamps were printed, the mechanical register decreased (Descending Register), blocking the press when the balance reached zero.

Attacks on Franking Systems and Bypass Methods

The transition to electronic systems and remote balance top‑up via telephone lines (modems) in the 1980s and 1990s gave rise to a whole range of vulnerabilities at the intersection of mechanics, circuit design, and cryptography.

Attacks on Resetting Internal Registers (Fiscal Memory)

Electronic postage meters recorded the balance in two independent types of non‑volatile memory (NVRAM): the main descending register (available balance) and the cumulative register (total amount spent over the entire lifetime).

• Voltage glitching attack: Attackers discovered that if, at the moment of a transaction (when the microcontroller subtracts the stamp cost from the NVRAM register), an artificial short‑term power supply drop is provoked (shorting the power line to ground or applying a high‑frequency pulse), the write logic to the memory cells is disrupted. The microcontroller managed to send the print command, but the power glitch prevented the reduced balance value from being correctly rewritten to EEPROM. As a result, the balance remained unchanged, while a legitimate stamp was printed.
• Physical cloning of memory chips: Attackers opened the protective cover, desoldered the NVRAM chips, read their contents on a programmer while the balance was at its maximum (e.g., £5000), and then reinstalled the chips via special sockets. When the balance ran low, they simply removed the chip and overwrote it with the previously saved dump, creating an endless cycle of free mailings.

Transition to IBI (Information‑Based Indicia) Cryptographic Stamps

To eliminate the threats of physical memory cloning, in the late 1990s the United States Postal Service (USPS), together with leading manufacturers (Pitney Bowes, Ascom Hasler), introduced the IBI standard. Instead of a plain‑text stamp, the machine began printing a high‑density two‑dimensional barcode (DataMatrix) containing a cryptographic signature.
Security architecture of the IBI stamp:

1. Stamp data: Unique machine serial number (Meter ID), date of mailing, exact postage rate, letter sequence number (transaction counter), and the sender’s and recipient’s postal codes.
2. Cryptographic core (Secret key): Inside the machine is a physical cryptographic processor (Hardware Security Module – HSM) protected against tampering to FIPS 140‑2 Level 3/4. It stores the machine’s private key SK_meter and computes a digital signature (MAC or ECDSA) over all the stamp data.
3. Flow control: Receiving terminals at sorting stations scan the barcode, automatically retrieve the machine’s public key PK_meter from a central database, and verify the signature’s validity.

Why Cryptography Did Not Fully Solve the Problem

A critical vulnerability ignored by the designers of IBI cryptographic stamps must be pointed out: asymmetry of operational risks and the problem of wholesale fraud.

• Replay Attack: An attacker can print one legitimate cryptographic stamp worth £2 for a letter to a specific recipient, and then run that stamp through a high‑quality duplicator onto 10,000 envelopes with the same address (e.g., mass advertising or bill mailing). From a cryptographic perspective, the signature on each of the 10,000 letters is absolutely valid.
• Economics of detection: To detect this duplicate, the sorting centre must check the uniqueness of the transaction counter for that specific franking machine in real time against a huge centralised database of all mail items in the country. If the database is unavailable or cannot handle the streaming load (millions of letters per hour), duplicates pass through the control system unnoticed.
• Insider collusion (Insider Threat): Large violators (mass marketing agencies) find it economically more profitable to bribe a shift supervisor or sorting‑node engineer to physically route their trucks with counterfeit stamps around the scanning lines directly to the loading area, rather than attempting to break the protected HSM chip inside the machine.

Additional Physical Protection Layers: ATMs, Communication Lines, and HSM Enclosures

To ensure business continuity and protect top‑tier assets, engineers must contend with aggressive attacks aimed at destroying the physical shell of devices or covertly intercepting signals from cable routes.

Physical Security of ATMs (ATM Security)

An ATM is a protected safe combined with a computer terminal, located in an open street or vestibule environment. The threat model for ATMs includes brute‑force destructive attacks:

• Gas attacks: Attackers pump an explosive gas (propane‑butane or acetylene with oxygen) through technical gaps (e.g., the cash dispenser slot) into the safe area, then remotely detonate the mixture with an electric igniter.

• Engineering countermeasure: Integration of Gas Suppression Systems. Specialised sensors detect the presence of combustible gas above a critical concentration and automatically spray a chemical inhibitor that neutralises the explosive properties of the mixture.

• Thermal and mechanical destruction: Use of disc cutters, hydraulic spreaders, or thermal lances that melt refractory steel by burning iron in a stream of pure oxygen at temperatures above 3000°C.

• Engineering countermeasure: Use of composite armour of the CEN (Composite Enhanced Safe) class. Safe walls consist of alternating layers of high‑strength steel, reinforced concrete with corundum chips (which instantly dull and destroy diamond discs), and polymer binders that, when heated, release huge volumes of choking smoke, making it impossible to continue work in the room.

• Intelligent Banknote Neutralisation Systems (IBNS): The point of stealing cash is completely destroyed if the notes are rendered unusable. Shock, tilt, pressure, and temperature sensors are installed inside the cash cassettes. When unauthorised opening or safe destruction is detected, pyrotechnic charges fire, instantly dousing the banknotes with indelible high‑contrast ink (blue, purple, or bright red). An ink‑stained banknote automatically loses its legal tender status.

Protection of Cable Routes and Communication Lines (Protected Distribution Systems – PDS)

Transmitting confidential traffic over copper or fibre‑optic cables outside the protected data centre is vulnerable to physical tapping.
To prevent unauthorised information interception, Protected Distribution Systems (PDS) are created:

• Overpressure monitoring: A copper cable is placed inside a sealed metal or plastic conduit into which dry air or nitrogen is pumped under pressure. Precision pressure sensors (manometers) are installed at both ends. If an attempt is made to drill into the conduit or make a cut to access the conductors, the pressure drops, which is instantly detected by the controller, disconnecting data transmission over that line until the response team arrives.
• Interference monitoring in fibre‑optic cables: Tapping a fibre‑optic cable is usually done by microbending, where part of the light flux begins to scatter through the cladding and is captured by a spy’s photodetector. To protect against this, engineers use specialised continuous fibre‑integrity monitoring systems. They measure phase shifts, signal attenuation, and Rayleigh backscattering using Optical Time‑Domain Reflectometry (OTDR). Any change in cable geometry or attempt to extract photons alters the interference pattern and triggers an alarm.

Hardware Protection of Servers and HSMs (Tamper‑Resistant Enclosures)

Top‑tier cryptographic processors (e.g., IBM 4758 or modern Hardware Security Modules) must guarantee that secret keys cannot be extracted even if the device falls into an attacker’s laboratory.

• Tamper Sensing Mesh: The processor core and memory chips are physically wrapped or coated with a flexible multilayer substrate on which a zigzag pattern of ultra‑thin conductors (tracks of chemically active polymer or carbon paste) is deposited with a pitch of less than a tenth of a millimetre. A low‑amplitude control signal is continuously passed through these tracks. Any attempt to drill the housing, lift the protective layer, or dissolve it with acid changes the resistance of the loop or breaks it.
• Potting compound: The entire computing module is potted in a solid opaque high‑density epoxy resin with added quartz sand. Attempts to mechanically chip or chemically dissolve this resin to access the chip pins lead to physical destruction of the silicon dies and the conductors leading to the substrate.
• Active zeroisation circuits: Mesh sensors, temperature sensors (detecting cooling with liquid nitrogen to prevent data leakage through residual magnetic effects – Cold Boot Attack), and X‑ray sensors are connected to non‑volatile static RAM (SRAM) where the root encryption keys are stored. This memory is continuously powered by a miniature lithium battery located inside the protective boundary. At the slightest breach of mesh integrity or if environmental parameters exceed operating limits, the SRAM power circuit is instantly shorted to ground (Crowbar circuit), causing complete physical erasure of the keys within nanoseconds – long before the attacker gains access to the dies.

Mechanical Fraud and Fiscal Control of Cash Registers

Control over cash turnover and resources requires metering systems to be resilient to specific attacks aimed at distorting historical metrics.

Mechanical and Inductive Coin Acceptors

Before the era of digital payments, vending machines, parking meters, and resource meters relied on mechanical coin verification. Primitive systems checked only physical dimensions (diameter and thickness) using calibrated slots and levers.

• Bypass vectors: Use of counterfeit tokens (slugs) machined from cheap scrap to the size of a legitimate coin, or the coin‑on‑a‑string technique. In the latter case, a thin, strong line was attached to a coin: the coin was dropped into the machine, activated the mechanical credit flag, and then carefully pulled back out.
• Inductive verification: Modern coin acceptors evaluate coins dynamically. The coin rolls down an inclined track past several electromagnetic coils. The flight speed (mass) and the change in inductance and Q‑factor of the coil loop are measured, allowing the electromagnetic conductivity of the alloy to be determined with high accuracy (distinguishing copper‑nickel alloys from pure steel or lead).

Fiscal Memory of Electronic Cash Registers

To prevent tax evasion by retail businesses (by deleting sales records from cash register memory at the end of the day), many countries have legally mandated manufacturers to implement Fiscal Memory.

• Architectural vulnerability of old cash registers: In early electronic cash registers, sales data was written to ordinary RAM or the hard drive of the controlling computer. Sellers used specialised software (Phantomware or Zapper) that modified the sales database just before sending reports to the tax authority, understating revenue by a fixed percentage.
• Engineering solution (Hardware Fiscal Storage): A modern cash register contains, inside a sealed housing, an independent non‑volatile module (in Russia – FN, in the EU – Fiscal Box). Writing to this module follows the WORM (Write Once, Read Many) principle – the memory architecture only allows appending new data blocks, but hardware‑blocks modification or deletion of existing sectors. Each fiscal operation is signed with a unique cryptographic code (cryptographic verification feature) generated by the built‑in cryptographic module. Altering records in the main register memory becomes useless: on reconciliation, the chain of cryptographic hashes in the fiscal storage would be broken, which is instantly detected by tax audit.

Energy Fraud at the Level of Mainline Systems

Resource metering manipulation occurs not only at the level of end‑user households, but also at the level of wholesale distribution markets. Let us describe cases of data manipulation (e.g., in the practice of the notorious Enron corporation).

• Manipulation of readings at mainline substations: In a deregulated electricity market, generating companies sell electricity to distribution networks through an electronic trading system based on current line loading. Attackers deliberately injected false data into SCADA telemetry systems. By simulating overload (artificially inflating current sensor readings on key transmission lines), they forced the grid automation to disconnect legitimate cheap power sources. This created an artificial local electricity shortage in large regions (e.g., California), allowing them to sell power from backup substations at astronomical speculative prices, causing billions in losses to the state and consumers.

Patterns of Vulnerabilities at the Intersection of Physical and Digital Worlds

A deep engineering analysis of the topics previously examined allows us to formulate key conceptual conclusions and systemic design errors, which we attribute to the category of Hardware‑Software co‑design flaws.

The Illusion of Absolute Cryptography

The most common mistake of young security engineers is the assumption that introducing a strong cryptographic algorithm automatically makes the system secure. The examples of KITAS tachographs and franking machines clearly demonstrate: attackers almost never attack mathematical primitives (AES, RSA, ECC). They attack:

• Analogue peripherals: Applying a magnetic field to the Hall sensor, jamming communication signals, thermal effects on PIR sensors.
• On‑board power: Voltage glitching to reset memory registers.
• Human factors: Bribing calibration technicians or inspectors, social engineering.

The Fail‑Safe vs. Fail‑Secure Logic Failure

In distributed systems, there is a fundamental conflict between system availability, human life safety, and asset protection.

Logic Concept	Behaviour on Failure / Attack	Application in Real Systems	Exploitation Vector by Attacker
Fail‑Safe	The system disables protection, priority is availability and life	Fire doors, ACS on evacuation routes, tachographs (issuing a communication error instead of engine lock)	Simulating a fire with aerosol, electromagnetic jamming of the motion sensor to force the device into error mode
Fail‑Secure	System Lockdown: full blocking of all functions and interfaces	Bank vaults, cryptographic HSM modules	Denial‑of‑Service (DoS) attack on a competitor’s business by deliberately triggering tamper sensors

An engineer must find a balance, being aware of how an attacker will deliberately drive the system into a technical failure mode.

Information Asymmetry and the Market for Lemons

Later in other parts, the crucial socio‑economic aspect of security will be raised: why do distributed protected systems often degrade?
When the end purchaser of equipment (e.g., a transport company buying a tachograph, or an energy consumer choosing a meter) is economically interested in concealing the true data, the security system manufacturer that covertly leaves backdoors or convenient loopholes for bypassing control wins in the market. Security becomes decorative if regulatory authorities do not conduct rigorous independent auditing and hardware reverse‑engineering of supplied devices.

Conclusion

Designing reliable distributed systems requires a Security Engineering specialist to understand the physics of materials, the laws of mechanics, circuit design, and microprocessor logic to the same extent as cryptographic protocols. The physical world has its own unique vulnerabilities and tolerances, and as long as digital bits are processed by real silicon transistors behind real steel doors, physical security will remain the foundational basis of trust in any information system.