GB2428357A - Generating secure keys at communicating parties by inferring the opening and closing of a switch at the other party from current flow observations - Google Patents

Abstract

Communicating parties X and Y (A) are connected via an electrical circuit with a power supply (B) and are each equipped with a switch (C). The parties operate their switches in a synchronised manner. In each switching cycle each party sets its switch open/closed in a first period and then reverses the switch position to closed/open in a second. Current flow is monitored throughout. When X and Y make the same switching choice (open then closed or vice versa) current will flow during one period, since this can be detected externally the result is discarded. If they make different switching choices no current will flow and it is not possible to detect externally which party made which choice. However X and Y can infer what the other did with its switch. Each switching option is assigned a bit value by the parties and thereby an encryption key can be built up. Also disclosed are systems to prevent detection of switch position by introducing noise and magnetic materials and related schemes using variable resistances and power supplies.

Description

Davide Antilli Propagation of deterministic and non-deterministic values

by means of electrical conductors System and method for the propagation of deterministic and non-deterministic values by means of electrical conductors

1 The Orlando System: background

1.1 State of the Art The secure transmission of a cipher key is fundamental to many cryptographic processes.

Altering the spin state of one of a pair of entangled' particles has been proposed as a method that would guarantee that a given transmission could not be intercepted or observed before it reached its legitimate destination. In principle this would allow an encryption key or even plain text to be transmitted without risk of compromising security but the difficulty of implementing the design has led engineers to develop alternatives, systems which exploit the fragility of information encoded as quantum states, or the perishable nature of a single photon. These systems work on the principle that interception or observation will be detected as an implicit characteristic of the received transmission: in the former case interception will corrupt' the energy state that embodies the message; in the latter, the energy comprising the state will be absorbed altogether.

1.2 Principle of the Orlando System: energy containment If no energy is released into the public domain communication may be considered private and secure. If, for example, Party X and Party Y occupy the same secure location - if, for example, they share a bunker - from a system perspective communication between X and V can be considered private, and secure.

Similarly, if Party X and Party V occupy private domains separated by some public domain, and if during the course of the communication no energy is released at either private domain into any part of the public domain, communication is also considered private, and secure. This is the principle upon which the security of the entangled particle' system rests; it is also the principle upon which the security of the Orlando system rests.

Matters of implementation and cost aside, Orlando has an advantage over the entangled particle' system. It would be technically possible for an unauthorised party to substitute an entangled mole' particle for the genuine article; the original particle, now in the secret possession of the third party, would intercept messages while the mole' were used to relay identical messages to the intended recipient, or vice versa. This is sometimes known as using a ringer', sometimes as an authentication issue'. Theoretically, all of the transmission systems mentioned here are vulnerable to ringers' - with the one exception of Orlando.

2.1 The system, "Orlando" The basic design of the Orlando system is shown in Figure 1. It consists of an electrical conductor (Figure 1, A') which forms a circuit that can be broken by opening switches (Figure 1, C') located at either end, X and Y'. The power supply (Figure 1, B') may be located anywhere in the circuit. 2 represents some unauthorised third party, assumed to be present at all points about the circuit that lie outside secure zones at X and V. The secure zones are shown in Figure 2, as D'. There may be a small power supply at both X and Y, or only at X, or only at V1 or at some neutral location, 0, within a third secure zone (Figure 2, 0'). X, V and 0 are all equipped with some means of detecting current, ie of detecting when the circuit is complete (Figure 2, E'). If X and V have their own power supply a connected circuit will instantaneously be detected as a back current'. X and V are equipped with a random digit generator (Figure 2, F').

F

Davide Antilli Propagation of deterministic and non-deterministic values by means of electrical conductors 2.2 Basic Inference (See Figure 1) The Orlando circuit provides information of two types: a) information gained directly from observation (flow of current) b) information gained by inference (state of switches inferred from current outcome) In some cases information may be gained by inference when no current flows, ie when no energy is directly transferred from any point in the circuit to any other point: In Figure la the switch at X is closed, the switch at V is open: no current flows. The observer at X infers that the switch at Y is open; the observer at V can make no inference, the switch at Y being open. The unauthorised third party observer Z can make no inference.

In Figure lb the switch at X is closed, the switch at V is closed: a current flows. The observer at X infers that the switch at Y is closed; the observer at Y infers that the switch at X is closed.

Z infers that both switches are closed.

In Figure ic the switch at X is open, the switch at V is open: no current flows. No party can make any inference.

In Figure ld the switch at x is open, the switch at V is closed. Y infers that the switch at X is open; X infers nothing, Z infers nothing.

In two out of the four cases, Figures la and id, either X or V is able to infer the state of the counterpart switch from the absence of a current. But this is not, properly speaking, information unless the counterpart switch operator knows that the inference has been made.

In example Figure la, X infers the position of the switch at V. If X does not inform Y that the inference has been made the inferred switch state is not information; but if X does inform V the information is not secure. X can only make the inference if the switch at X is closed; if the switch at X is closed, and no current flows, then the switch at Y must be open. Any third party observer who overheard X confirming the inference would learn the position of both switches.

2.3 Secure method We will consider the observations and inferences for a single special case. The system and method for this special case will allow X and V to share information equivalent to a single binary digit in a secure manner. It follows that a series of q occurrences of this special case will allow X and Y to share information equivalent to q binary digits in a secure manner. The q digits obtained will be a random binary series which can be used as an encryption key.

Method 1. X and V each choose at random, and do not disclose, either the digit or the digit 0'.

2. X and V expand their single-digit value into a 2-digit value as follows: 1 = 10; 0 = 01. This will be known as a Switching Pair.

3. The Switching Pair represents switching instructions. Let the digit 1 represent the instruction, Close your switch' and the digit 0' represent the instruction, Open your switch'.

4. Switching is performed in pairs of iterations. The Switching Pair 10' represents the instruction, For the first iteration, close your switch; for the second, open it'.

5. X and Y set their switches in the positions indicated by their respective Switching Pairs one iteration at a time; current is applied to the circuit, and the outcome noted.

Davide Antilli Propagation of deterministic and non-deterministic values by means of electrical conductors 6. If, for example, X and Y both happen to pick the Switching Pair OV, the following will occur.

For the first iteration (Figure ic) both X and Y will open their switches; power will be applied and the absence of a current noted. The first iteration ends here. For the second iteration (Figure ib) X and V will both close their switches, power will be applied, and the flow of current noted, which will complete the second iteration.

The protocol given below allows X and V to infer the state of their respective counterpart switches in every instance. It will be seen that when X and Y happen to pick different binary values the information gained by inference is secure. When X and V pick identical binary values the information gained will not be secure as some third party observer Z will be able to make identical inferences. In this case the information gained by inference is discarded.

2.4 Secure protocol: non-secure Switching Pair combinations (Figures lb and ic) If X and Y happen to pick the same Switching Pair (SP) the following will occur:

Example 1

SP Iteration 1 Iteration 2 X (01) open closed V (01) open closed Outcome no current current Both X and V had their switches open for the first iteration and no current flowed when the power was applied.

For the second, both switches were closed, and a current flowed. Since a current will only flow if both switches are closed the switch positions for X and V will be evident to any observer of the current/no current outcome, ie some third party Z will be able to infer all that is inferred at X and V. If one of the positions in a Switching Pair is known, the second is given, since only the combinations (closed, open) and (open, closed) are allowed. The switch positions at X and V for both iterations will therefore be known to any observer of the current/no current outcome.

The combinations (X=1O, V=10); (X=01, V=01) are logically equivalent.

2.5 Secure cases: non-matching pairs (See Figures la and id.) Example 2 shows one of the two possible non-matching Switching Pair combinations:

Example 2

SP Iteration 1 Iteration 2 X (10) closed open V (01) open closed Outcome no current no current 2.51 Observation and inference at X After Iteration 1, X infers that the switch at V must have been open, or a current would have flowed (Figure la). X can also infer at this point that the switch at V will be closed for Iteration 2: V cannot open the switch for both iterations; the switch at V was open for the first iteration, therefore it will be closed for the second. X knows Vs switching pair to be the combination (0,1).

Davide Antilli Propagation of deterministic and non-deterministic values by means of electrical conductors 2.52 Observation and inference at Y Ys switch was open for the first iteration allowing no inference may be made as to the switch state at X (the case is identical to that of X in Figure id). For the second iteration, however, the closed switch at Y allows Y to infer from the lack of a current that the switch at X was open (Figure la); this also gives the position at X for the first iteration as closed, since the sequence open, opens is not allowed. Y knows Xs switching pair to be (1,0).

2.53 Observation and inference at Z Any third party, Z, familiar with the Orlando protocol and the exact synchronisation procedure, will know that in each of the iterations in Example 2 the circuit was broken by a single open switch. Since the current7'no current' outcomes i) (X=0, Y=1) ii) (X=1, Y=O) are equivalent and interchangeable, Z will not be able to infer with of the two switches was open and which closed for any given iteration. The information, "X's switching pair was the combination (1,0') ", known to X and Y, will not be known to Z.

2.6 Summary of method

Where the Switching Pairs at X and V match energy is released and the process is not secure.

The information gained from such instances is therefore discarded.

Where the Switching Pairs at X and V do not match no current will flow in either of the Switching Pair iterations, le no energy is released by either X or V into the intervening medium. The lack of a flow of energy is nonetheless information: it is sufficient information to allow X and Y to infer the counterpart switch states for both iterations. However, the lack of a flow of energy from either X or Y into the intervening medium qualifies the process as secure from any attempt by any third party Z to derive any knowledge of the state of the switches at either X or V for either iteration.

To complete the process X and V agree to take one of the two values, the switch state at X or the switch state at V. as the secure value. When sufficient secure values have been obtained the string can be used as a one-off encryption key.

3 Authentication An agreement between X and Y to begin Orlando communication at fixed times, eg every day at 17.00 hours, will be sufficient to guarantee transmitter/receiver authentication. Some hostile third party could seize control of the intervening conductor and pose as Y vis a vis X and, simultaneously, as X vis a vis Y; but the Orlando strings obtained by X and V would not match, and standard routine processes would ensure that the imposition were detected.

Suppose Z attempted to duplicate the Orlando string obtained with X while posing as V. If this secure string required L iterations, it would take 2L iterations to ensure that Y obtained the same Orlando string. Again, obvious routine processes would ensure detection.

4 Magnetic effects The conductive medium between X and V may be susceptible to magnetisation. The critical switching positions, ie (X=open; Y=closed, and vice versa) may form a kind of horseshoe magnet if the conductive medium connecting the switches were magnetised, with flux density greatest around the poles, ie the open switch. Magnetising the circuit and measuring flux Davide Antilli Propagation of deterministic and non-deterministic values by means of electrical conductors density at several points would therefore allow some unauthorised third party Z to determine which of the two switches, X or Y, was open and which closed.

4.11 Figure 3 shows one way of forestalling the threat. Within the secure zones at X and Y contrasting ferromagnetic (G) and diamagnetic conductors (RH) (eg iron; copper) are placed in series and a diamagnetic switching mechanism (I1"; 12) set perpendicularly to the two ferromagnetic conductor elements.

For the sake of completeness the additional steps 4.12 - 4.17 are given below.

4.12 Since electrical conductivity and magnetic permeability are not equivalent, and conductive and non-conductive elements manifest similar magnetic properties, randomly and constantly modifying the internal switching arrangement of the switching mechanism to vary its magnetic flux output (Figure 3, Ii and 12) will reduce the probability of producing an instance of statistically identifiable electrical conductivity to zero, even at infinitesimal distances.

The switching mechanism.1, might be moved or vibrated and might, but need not, consist of randomly switched circuits of magnetically and electrically variegated elements: copper, gold, iron, silicon, air, vaccuum etc. 4.13 Figure 4 shows an alternative method to that proposed in 4.11. Part, or all, of the conductor (Jr) is magnetised, and the switch mechanism (K') given the contrasting polarity.

Some ferromagnetic conductor such as iron would be used as the closed' switch, a ferromagnetic non-conductor such as ferrite or stainless steel as the open' switch; in the latter case, the ferromagnetic material need not of course be brought into actual contact with the conductor.

If the magnetic flux density of the switches is randomly varied (as may very easily be done) continuous close observation will not allow Z to determine the electrical conductivity of the switching mechanism in any given instance.

4.14 An additional alternative to the method proposed in 4.11 is shown in Figure 5. Where the intervening conductive medium is locallly or predominantly ferromagnetic (L) an electrical non-conductor of high magnetic permeability (M) such as ferrite, can be placed at either end of the circuit to create a permanent donut' magnetic circuit.

4.15 Magnetic noise generated at X and V will obliterate any variations in magnetic effects.

4.16 Magnetic shields can be used to prevent stray magnetic flux from leaving the secure zones at X and V. 4.17 Part of all of intervening conductive medium, as well as the switching mechanism, can be demagnetised using solenoids or any other appropriate method.

Variations on the system and method: preliminary: Three methods of sending deterministic values across electrical conductors are described here: all of relevance as each forms the basis of a variation of the Orlando secure method.

5.1 The transmitter X transmits signals to the receiver V by opening and closing the switch located at X according to a pre-agreed protocol, eg Morse Code. Closing the switch at X results in a electrical activity at Y; opening it at X results in the cessation of electrical activity at Y. Nb The sole power supply for the circuit might be located exclusively at the transmitter Y. The transmitter X expends energy opening and closing his switch but is not necessary for any of the energy expended at X to reach V for the transmission to be successful.

5.2 The transmitter X transmits signals to the receiver V by varying the input of a power Davide Antilli Propagation of deterministic and nondeterministic values by means of electrical conductors supply located at X. Readings are taken at Y. The power input at X is information.

5.3 The transmitter X transmits signals to the receiver Y by varying the resistance of the X-Y circuit by means of a variable resistance (a slider') located at X. (See Figure 5, P1 and P2'). The total circuit resistance readings are taken at V. The incremental circuit resistance introduced at X is information.

5.4 Variations on the secure method 5.41 Varied power input at X and V 1. X and V are each equipped with a variable power supply unit, and a switch, all in series: power supply units have n settings, so that X and Y can choose to apply one of n different power levels to the circuit. If the serial step increases in power (0, 1, 2... n) are made sufficiently large, loss of power over the circuit will not affect accurate measurement.

2. X and Y choose, but do not disclose, a power level, eg X chooses (n-v), y chooses (n-w).

3. X and Y simultaneously close their switches and measure the current 4. X and Y will both measure the level (2n-(v+w)). X obtains the value (nw); y obtains (n-v).

Any unauthorised third party Z will obtain nothing beyond the combined output (2n-(v+w)).

5. X and V can agree in advance which of the two values, v or w, to take as the secure value; or they might take both, as discrete values in series, eg (v,w).

5.42 Varied resisthnce at X and V 1. X and Y are each equipped with a variable resistance, ie X and V can choose from a number of resistances to place in series or in parallel within the basic XV circuit. Variable resistors of this type are sometimes known as sliders' (Figure 5, P1' and P2) 2. X and Y choose, but do not disclose, resistances Rx and Ry respectively.

3. X and Y simultaneously close their switches; a known voltage is applied to the circuit, and the total circuit resistance (Rx + Ry) is found by applying the law V = IR.

4. X and Y work out the value of their respective counterpart's resistance from the value (Rx + Ry) (which may be disclosed in the public domain) and their respective values Vx and Vy (which may not be disclosed).

5. X andY can agree in advance which of the two values, Rx or Ry, to take as the secure value; or they might take both, as discrete values in series, eg (Rx, Ry).

5.43 General case Any property of an electrical circuit, or of a magnetic circuit, eg flux density, whose gross effect is the combined and instantaneous outcome of changes effected by X acting independently at X and by V acting independently at V, may be taken as the basis for secure transmission by means of Steps 2, 3 and 4 described in 5.41 and 5.42 above.

6 Varied resistance at X and V If some third party Z attempted to determine which of the two switches was closed by placing some resistance in parallel with the Orlando circuit and measuring total circuit resistance, it would infallibly be detected in two ways: a) X (and Y) would note a current' outcome whenever the switch at X (or Y) was closed Davide Antilli Propagation of deterministic and non-deterministic values by means of electrical conductors b) the values obtained at X and Y would not correspond - the communication would be grossly corrupted Suppose Z found some means of creating a permanent circuit of a kind such that, when the switch at X was closed and the switch at Y open (and vice versa), the instantaneous electrical activity (ie back current) at X's power supply were too small for X to notice. Suppose, too, that when both switches were closed the instantaneous electrical activity at X and Y occurred at the expected levels and the imposition remained undetected. If this were possible, comparison of total circuit resistance when X and Y supposed no current to flow would allow Z to determine which of the two switches was closed.

However the data collected by Z would be valueless if X and V were equipped with variable resistance sliders' (Figure 5, P1' and P2') and the resistance contributed by the portions of the sub-circuit encompassing, respectively, X and Y, constantly and randomly varied.

7 Synchronising device X and V can synchronise their switching by means of a separate circuit. X or V or some other party 0 opens a switch placed in the parallel separate circuit. X and Y each have a separate power supply placed in series in the parallel circuit. When the switch is closed a back current occurs simultaneously at the power supplies at X and V. This back current triggers some other event, eg starting/stopping a clock, changing the position of a switch, modifying the level of a power input device into a separate circuit etc. 8 Disruption The Orlando system would be relatively easy to disrupt: some hostile third party could easily break the circuit, either permanently or by means of a switch.

There are, however, a very great number of available electrical circuits between any two points. The PSTN and the National Power Grid might each provide several alternative routes between X and V. It would be relatively easy to establish a centralised routing system which would find alternative circuits if one were found to be defective.

9 Parallel communication channels It might be useful to establish a parallel communication channel - this need not be secure - for the purposes of synchronisation, authentification and establishing the need for alternative circuit paths: a mobile or fixed telephone connection, for example; a fax machine, or any available telecommunications channel.

Key to diagram component lettering A Diamagnetic X-Y electrical conductor B Power supply (fixed or variable output) C Switching mechanism D Secure zones E Current gauge F Random binary value generator G Contrasting ferromagnetic electrical conductor H Contrasting diamagnetic electrical conductor Ia One-off conductive switch setting of variable magnetic permeability lb One-off non-conductive switch setting of variable magnetic permeability 3 Magnetised portion of electrical conductor running between X and Y K Magnetised switch of contrasting polarity to 3 L Ferromagnetic X-Y electrical conductor M Ferromagnetic non-conductor N North pole of a magnet 0 Neutral secure zone P1 Fixed (ie X-Y conductor component) portion of variable resistance slider' P2 Moveable (le switch component) portion of variable resistance sIider S South pole of a magnet Davide Antilli System and method for the propagation of deterministic and non-deterministic values by means of electrical conductors

Claims (1)

1. GBO5 14220.3

   I A physical system for generating a secure cipher key consisting of an electrical circuit connecting independently controlled switches located at either end X and Y of the circuit and operated by Operators X and Y respectively.

   2 A method whereby the successive settings of the switches claimed in Claim 1 are regulated by the random choice of a binary variable, choices which are not disclosed by Operator X to Operator Y or vice versa, or to any other party.

   3 A further step within the method claimed in Claim 2 whereby the physical regulation of the switches at X and Y is performed in synchronised iteration pairs, ie where Operator X and Operator Y simultaneously set their respective switches according to the method claimed in Claim 2 and note the current/no current outcome; and subsequently simultaneously reset their respective switches and note the second and final current/no current outcome.

   4 A further step within the method claimed in Claim 2 whereby the switch setting for the second of the iterations of a pair is always the counterpart of the first, ie if the setting "open" is chosen for the first iteration, the setting "closed" is chosen for the second, and vice versa.

   A step within the method claimed in Claim 2 whereby Operators X and Y, having performed a switching iteration pair as claimed in Claims 2-4, logically infer the positions of their respective counterpart's switches from the flow (or lack of flow) of current in the first and second iterations.

   6 A step within the method claimed in Claim 2 whereby Operators X and Y discard the results where a flow of current in the first or second iteration of an iteration pair indicates that they have chosen identical or "matching" pairs.

   7 A step within the method claimed in Claim 2 whereby Operators X and Y retain the results where the absence of a current in both the first and the second iterations of an iteration pair indicates that they have chosen different or "non-matching" pairs.

   8 A step within the method claimed in Claim 2 whereby Operators X and Y agree to take the initial binary value of a given Operator's performed iteration pair as the Secure Digit, eg Operators X and Y might agree that, where Operator X's first switch setting in a given iteration pair is respectively known or inferred to have been "closed", the digit "1" will be added to the encryption key string recorded by Operators X and Y. 9 Further steps within the method claimed in Claim 2 whereby many iteration pairs are performed by X and Y as may be required to build a cipher key string of a given length by means of the step claimed in Claim 8.

   A step within the method claimed in Claim 2 whereby Operators X and Y agree to begin performing iteration pairs at some precisely defined and independently verifiable time, which time need not to be agreed via a "secure" communication.

   11 A physical system as claimed in Claim 1 where Operators X and Y are fully automated entities.

   12 A physical system as claimed in Claim 1 where Operator X and Y are manual operators assisted by appliances such as clocks, random number generators etc. 13 Switches as claimed in Claim 1 where the resistances of the switch settings at X and Y are independently and randomly varied, ie where Operator X does not disclose in advance to Y the resistance value of the switch at X to be deployed for any given switching iteration, and vice versa.

   14 An electrical circuit as claimed in Claim 1 where Operators X and Y each control an independent power supply of variable output, and where the output is randomly varied without prior disclosure to any second party, ie where Operator X does not disclose in advance to Y the voltage to be applied at X for any given switching iteration, and vice versa.

   An electrical circuit as claimed in Claim 1 where the physical elements used for the composition of the switches at X and Y and for the portions of the intervening conductor lying within some secure domain at X and Y respectively are composed of materials of contrasting magnetic properties (ferromagnetic, diamagnetic etc) and their physical design not disclosed to any second party by Operators X and Y respectively, ie where the design of the switch at X is unknown at Y and vice versa.

   16 An electrical circuit as claimed in Claim 1 where Operators X and Y create random electrical "noise", ie where X and Y induce random currents along the portion of the conductor that hosts the switch, eg by earthing the conductor, by allowing an electrical charge to flow in the vicinity of or around an open switch, or by any other means that will produce electrical activity to disguise or screen the state of the switch.

   17 An electrical circuit as claimed in Claim 1 along which X and Y create random magnetic "noise", ie where X and Y induce random magnetic flux along the portion of the conductor that hosts the switch.

   18 A system and method as claimed in Claims 1-17 where the current/no current outcome is observed by some co-operating third party, and the outcome communicated to X and Y by some "insecure" means, for example over a telephone line.

   19 A system and method as claimed in Claims 1-18 where X and Y open a separate "non-secure" channel in parallel in order to compare measurements of current, resistance and voltage, exchange information about their respective power/resistance settings for completed iterations, to compare timing, synchronisation etc in order to detect possible interference by some unknown third party Z. An electrical circuit as claimed in Claim 1 whose power supply is located at only one of X and Y. 21 An electrical circuit as claimed in Claim I whose power supply is located at both XandY.

   22 An electrical circuit as claimed in Claim 1 whose switches consist of randomly or non-randomly variable resistances whose value represent numerical values after the fashion of a modulated signal.

   23 A step whereby Operators X and Y at either end of a communications channel as claimed in Claim 22 simultaneously set their respective resistances to their respective chosen values in a synchronised fashion and apply a known voltage to the circuit.

   24 A step whereby Operators X and Y measure the total circuit resistance after performing the step claimed in Claim 23 and calculate the resistance introduced into the circuit by their respective counterpart and so derive the intended numerical value.

   An electrical circuit as claimed in Claim I whose switches consist of fixed resistances of known value and to which power is applied independently at both X and Y. 26 A step whereby Operators X and Y simultaneously modulate the output of their respective power supplies as claimed in Claim 25 either as a continuum or as a discrete series, randomly or otherwise, which modulations represent numerical values after the fashion of a modulated signal.

   27 A step within the method claimed in Claim 25 and 26 whereby Operators X and Y calculate the voltage introduced into the circuit by their respective counterpart and so derive the intended numerical value.