GB2380369A

GB2380369A - Encryption system for a communication network

Info

Publication number: GB2380369A
Application number: GB0123231A
Authority: GB
Inventors: John Martin Williams; Michael Kitchin
Original assignee: Marconi Communications Ltd; Marconi Co Ltd
Current assignee: Marconi Communications Ltd; BAE Systems Electronics Ltd
Priority date: 2001-09-27
Filing date: 2001-09-27
Publication date: 2003-04-02
Anticipated expiration: 2021-09-27
Also published as: GB0123231D0; GB2380369B

Abstract

A communication network comprises a plurality of radio terminals 1 each having an encryption key controller 3. An encryption system used with the network has an allocated number of encryption keys, said number being one more than the number of terminals in the network. Each terminal is allocated all but one of the allocated number of encryption keys so that each terminal has a different omitted encryption key. Encryption is carried out using a key which is common to all of the terminals and if one of the terminals is stolen, the key omitted from the stolen terminal may be used to update the remaining terminals. In this manner the network may be kept secure even if a large number of terminals are stolen or lost over a period of time.

Description

A COMMUNICATION NETWORK ENCRYPTION SYSTEM The present invention relates to an encryption system for use with a communications network. Such a network may be a portable secure radio network, a telephone network or a computer network for example.

According to the present invention there is provided an encryption system for a communication network, the communication network comprising a number of terminals, the encryption system having an allocated number of encryption keys, the allocated number being one greater than the number of terminals, wherein each terminal is allocated a number of the encryption keys one less than the allocated number and contained within a key controller and each terminal has omitted therefrom a different encryption key from each other terminal, encryption being carried out within the communication network using a one encryption key common to all the terminals.

The present invention will be now described by way of example, with reference to the accompanying drawings in which :Figure 1 illustrates a communications network having five users; and Figure 2 shows a diagrammatic view of a User Terminal from Figure 1 This particular example is of an encryption system for use in a portable radio network, such as a police radio system or a military radio system, where typically some or all of the following circumstances prevail: 'A number of radios share the same key-a radio net is in operation

'The probability of losing radios and their keys is high 'It must be impossible to recover the net key from the contents of a stolen radio 'Power, space and computing power is strictly limited

'Keys are loaded extremely infrequently 'Anti-tamper precautions are undesirable 'A'panic erase'button is undesirable 'Methods of excluding a radio from the network should resist subversion at- tempts.

'The number of switches and controllers on the radio must be minimised 'Any operator intervention should be simple and intuitively convincing These requirements are common to virtually all secure portable radios.

The basic idea is to use a number of keys, allocated to the net members in such a way that the choice of key can be used to exclude any member of the group. A key updating using the key not possessed by a lost radio and carried out upon its loss prevents derivation of the net key from the contents of stolen radios.

The system is best described by a small example. Assume five users, 1, A to E, are connected by connections 4, to form a net. They are initially issued with keys kO to k5 held in a key controller 3. Each one of these keys, kO to k5, is usable as a key in its

own right. The allocation of these keys to A to E is as follows : User A receives kO, k2, k3, k4, and k5 (but not kl) User B receives kO, kl, k3, k4, and k5 (but not k2) User C receives kO, kl, k2, k4, and k5 (but not k3) User D receives kO, kl, k2, k3, and k5 (but not k4) User E receives kO, kl, k2, k3, and k4 (but not k5)

Initially, all the Users use kO. Assume, now, that one of the Users loses his radio, say user A loses his radio. The remaining net members would then switch to kl, the only key not possessed by User A, and use it update all their held keys. So kO to k5 would become k'O to k'5 and k'l would be used as the net key.

Assume, then, that at some later time another radio is lost, say D's radio is stolen. All users would then use k'4 (not held by D) to update all their keys again, so that k'0 to k'S would become k"O to k"5 respectively and k"4 would become the net key.

Provided updating can be carried out immediately a radio is lost, before another is lost, it is obvious that this system is capable of tolerating loss of virtually all the radios and that even the combined keys recovered from all the stolen radios cannot be used to derive the net key.

Operationally, the Users simply have to indicate to their radios the identity of the lost radio for the keys to update automatically under control of the internal microprocessor. This task could be automated and the instructions to do it sent over the air from a central point. The updated form of kO would be a convenient key to use for a dedicated update instruction channel; this is available to everyone who should have it. The instruction need not be kept private and can be sent over an open channel, although authentication will then be needed.

Depending on the size of the net, the voice quality, and operational constraints, au-

thentication may be done by voice recognition. Electronic certification methods may be needed in other cases however, and certainly would be needed in the case of automated systems.

It should be noted that this processor is not required to perform exponentiation or any other numerically intensive computation and can be of very modest power. Key updating operations are limited to passing old keys through a crypto chip to produce the new keys. The network update will take microseconds rather than milliseconds.

Note also that there is no need for anti tamper precautions, a panic erase switch, nor stun/kill facilities, and hence no reason why the keys should not be stored in inexpensive nonvolatile memory.

In order to introduce a new (replacement) radio to an existing net, it will be necessary for the key loading facility to have performed the same updates in the same order as has occurred in the net, so that the keys to be loaded into the new radio will match the old.

The advantages of this scheme are as follows:

'It is a practical solution to nets where loss of radios is frequent.

'Its processing requirements are extremely modest 'Exclusion of radios from the network cannot be prevented by their holders.

'Stun and kill arrangements are unnecessary.

'No need for emergency erase & battery backed RAM system.

'Re-keying a net can be an infrequent event 'Operator controls are minimised.

'Intuitively easy to understand & operate 'Updating may be done automatically from a central point.

Scalability The basic constraints on the size of the net are the rate at which radios are lost, and the ability to react to loss quickly and disseminate the updating instructions.

If the rate of loss is such that it becomes feasible for a cryptographically sophisticated attacker to steal two or more radios before either can be updated, then the network becomes compromised. This is because the keys they contain can be extracted and a complete key set assembled.

Estimating the size of a net should therefore be made by taking into account: 'The sophistication (or otherwise) of the potential attacker 'The rate of loss of radios 'The time taken to reconfigure the network to exclude radios.

Although the system requires the storage of much larger amounts of key information than is usual, this is not a major constraint due to the enormous capabilities of modem storage chips.

The process of updating keys is extremely simple in the case of block ciphers, and consists of a single encrypt operation. Thus if it is intended to update all the stored keys using kl, kl would be used as the key for the crypto and encrypts each stored key, including kl, to form encrypted versions-k'O to k'6. These would then replace the original kO to k6. The process is inherently one-way and once kO to k6 have been replaced by k'O to k'6, it is not possible to go back and work out kO to k6.

It should be noted that updating operations do not commute with each other. If k'4 is subsequently used to update all the stored keys, then the results, k"O to k"6 would be different from the k"O to k"6 had k4 been used first followed by k'l.

Claims

CLAIMS 1. An encryption system for a communication network, the communication net- work comprising a number of terminals each having a key controller, the en- cryption system having an allocated number of encryption keys, the allocated number being one greater than the number of terminals, wherein each terminal is allocated a number of the encryption keys one less than the allocated num- ber and contained within a key controller and each terminal has omitted there- from a different encryption key from each other terminal, encryption being carried out within the communication network using the one encryption key common to all the terminals.
2. An encryption system as claimed in Claim 1, wherein if a terminal is removed from the communication network, the remaining terminals using their key controller, update their encryption keys and continue encryption using an up- dated encryption key common to all the remaining terminals.
3. An encryption system substantially as hereinbefore described, with reference to and as illustrated in the accompanying drawings.