Patent Application
20150143126
The present invention relates to a method and apparatus for establishing a security association between a client terminal and a service node in order to deliver a push-type service and in particular, though not necessarily, to such a method and apparatus which employs a Generic Bootstrapping Architecture.
In order to facilitate the provision of services to user terminals, a mobile network such as a 3G network will often require the establishment of a secure communication channel or “security association” between client terminals (i.e. mobile terminals) and the network-based service nodes which provide the services. The Generic Bootstrapping Architecture (GBA) is discussed in the 3GPP Technical Specification TS 33.220 and provides a mechanism whereby a client terminal (UE) can be authenticated to a Network Authentication Function (the service node), and secure session keys obtained for use between the client terminal and the Network Authentication Function. The simple network model for this architecture is illustrated in
According to the GBA mechanism, a UE initiates the key generation process by sending a request containing a user identity to the BSF. The request also contains the identity of the NAF. The BSF retrieves an authentication vector from the Home Subscriber System (HSS), each authentication vector consisting of a random number RAND, an expected response XRES, a cipher key CK, an integrity key IK and an authentication token AUTN. The BSF generates key material KS by concatenating CK and IK contained within the authentication vector. The BSF generates a key identifier B-TID in the format of a NAI by base64 encoding the RAND value and combining the encoded value with the BSF server name, i.e. as
base64encode(RAND)@BSF_servers_domain_name.
The BSF retains the key KS in association with the transaction identifier B-TID and the NAF identity. The B-TID and AUTN are sent by the BSF to the UE, the USIM of the client terminal verifying the value AUTN using the shared secret K and returning a digest of the expected result XRES to the BSF. The USIM also generates the key material KS using the secret K and the value RAND (recovered from the B-TID).
Following completion of this procedure, the UE communicates to the NAF, the received B-TID. The NAF and the BSF are authenticated to one another, and the NAF sends to the BSF the received B-TID together with its own identity. The BSF uses the B-TID and the identity of the NAF to locate the correct key KS, and uses KS to generate a NAF key. Other information such as the NAF identity is also used in the generation of the NAF key. The generated NAF key is returned to the NAF. The UE is similarly able to generate the NAF key using the key KS that it has already generated.
After the GBA mechanism has been run for the first time, subsequent requests to establish a security association between the UE and the same or a different NAF may use the already established key material KS, providing that key has not expired. However, this will still require that the UE initiate a request for establishment of a security association by sending its B-TID to the NAF.
There are occasions on which it is desirable to allow the NAF to initiate the establishment of a security association with the UE. For example, one might consider a push-type service, which delivers news, sports, and financial, etc information to users who have previously registered for a service. A typical operational procedure to achieve this might be for the service provider to send an SMS message to the UE which requests the user to open a secure connection. However, there are many threats related to this model as an SMS might be manipulated, sent by an unauthorized party, be replayed, etc. If a security association existed, or the service node could initiate one, before the actual service data is sent, security procedures could be based on this and most problems could be mitigated.
According to a first aspect of the present invention there is a provided method of establishing a security association between a first node and a second node for the purpose of pushing information from the first node to the second node, where the second node and a key generation function share a base secret, the method comprising:
It will be appreciated that the key generation function may be a stand-alone node or may be a distributed server. In the case of a 3G network employing the Generic Bootstrapping Architecture, a Bootstrapping Server Function and a Home Subscriber Server may together provide the key generation function, where the Bootstrapping Server Function communicates with the service node and with the Home Subscriber Server. In the case of a 2G network, the key generation function may be a combination of a Bootstrapping Server Function and an AuC server.
In the case of a 3G network employing the Generic Bootstrapping Architecture, the service node comprises a Network Application Function. The step of generating a service key at the key generation function comprises the steps of:
The step of generating the service key at the client also comprises these two steps.
Said step of generating a service key at the key server may utilise values other than those sent to the client by the service node. The client may obtain certain of those other values from the key server.
Said additional information may comprise one or more of:
Said additional information may comprise a transaction identifier in the format of an NAI, and comprising an encoded random value.
Said additional information may be forwarded from the service node to the client in a message also containing service data, the service data being encrypted with the service key, wherein the client can decrypt the encrypted data once it has generated the service key.
In one embodiment of the invention, the key generation function sends to the service node a network authentication value. The service node forwards this value to the client, together with said additional information. The client uses the base secret and the authentication value to authenticate the key generation function. Only if the key generation function is authenticated does the client generate and use the service key.
In an alternative embodiment of the invention, the client requests an authentication value from the key generation function after it has received said additional information from the service node. Only when the client has authenticated the key generation function is the service key generated and used.
The terminal may comprise means for receiving from the service node a message authentication code, the terminal comprising means for generating an authentication key or keys from at least a part of the key generation information, and using the authentication key(s) to authenticate the message authentication code. The generation means may be a USIM/ISIM.
Said service key may be a Diffie-Hellman key for the second node, the method further comprising the step of providing to the first node a Diffie-Hellman key for that first node, and sending the Diffie-hellman key for the first node to the second node, said security association being established on the basis of the two Diffie-Hellman keys.
According to a second aspect of the present invention there is provided a service node for delivering a push service to a client via a secure communication link, the service node comprising:
In the case of the Generic Bootstrapping Architecture, said additional information comprises a B-TID containing the RAND value. Said means for forwarding is also arranged to forward to the client an identity of the service node.
According to a third aspect of the invention there is provided a client terminal for receiving a pushed service delivered by a service node, the client terminal comprising:
According to a fourth aspect of the present invention there is provided a key generation function for use in establishing a security association between a client and a service node for the purpose of pushing information from the service node to the client, the key server comprising:
According to a fifth aspect of the present invention there is provided a method of establishing a security association between first and second clients for the purpose of pushing information from the first client to the second client, where the first and second clients have trust relationships with first and second key servers respectively and share a secret with their respective key servers, the method comprising:
According to a sixth aspect of the present invention there is provided a method of protecting a node against replay attacks, the method comprising:
Embodiments of this aspect of the invention allow the second node to reject replay attacks based upon messages previously sent to the second node in respect of a valid GBA procedure. If the attacker were to merely increment that replay prevention value to a previously unused value, the second node would detect this change based upon the incorrect MAC value, and would hence detect the attack. Again, the first node may be a NAF server, with the second node being a client, or both the first and second nodes may be clients. It will be appreciated that features of the first to fifth aspects of the present invention may be combined with those of the sixth aspect, and vice versa.
The general Generic Bootstrapping Architecture (GBA) for 3G networks has been described with reference to
This discussion concerns the provision of a push service to a client. Typically, the client will have pre-registered with the service provider, but the initiative to push particular information is taken by the service provider. In such a situation, the service provider and the client will not already have a security association established with each other (security associations are typically short-lived), and one must be established.
A first solution proposed here takes the approach that the NAF asks the BSF for a NAF (or service) key. The BSF returns to the NAF, the NAF key together with the client transaction identifier (B-TID) and the corresponding network authentication value (AUTN). As has been stated above, the B-TID contains the encoded RAND value (as the NAI prefix), which can be used by the client to derive the base key (KS). The NAF can now compose a message containing the B-TID, AUTN, and further data including the NAF identity that the client requires in order to derive the NAF key, and send this message to the client. This message can be a message that only triggers the set-up of a SA (i.e. sharing of a service key) or it could contain service data (i.e. payload data) encrypted with the service key. In both cases, the values B-TID, AUTN, and other data required by the client to generate KS are sent in plain text but are “signed” with a Message Authentication Code. Note that the key(s) in the SA are derived using the key shared between the HSS and the UE, and that the AUTN is included in the message. It is therefore not possible to “spoof” messages even though the key used for integrity protecting the message is derived from the very SA it is intended to establish.
When the client receives the message, it retrieves the RAND part of the B-TID (by reversing the encoding) and the AUTN and applies them to the USIM/ISIM to derive the base key Ks. Then it uses the further data to derive the NAF key, and verifies the received message using the MAC.
The signalling exchanges associated with this procedure are illustrated in
In order to prevent the manipulation of the further data (required by the client) by the NAF, the BSF may sign that data using a derivative of KS. This may be important, for example, to prevent the NAF from extending the lifetime of a key.
The solution presented above allows the NAF to push to the client the information required to establish a security association between the two parties. Thus the client does not have to set up a connection with the BSF to perform these tasks. This represents an extremely time efficient solution. However, it requires that the NAF relay all key related information (key lifetime, Add-info, etc) in a protected form from the BSF to the UE. The B-TID and the other data might then comprise quit a large data structure. This might be problematic in the case where the volume of data that can be incorporated into the message structure used between the client and the NAF, e.g. where this structure is SMS.
In order to reduce the required data volume exchanged between the NAF and the client to establish the security association, the above solution may be modified by omitting the AUTN value from the data sent by the BSF to the NAF. The NAF now composes a message containing the B-TID and other necessary data (including the NAF identity) that the terminal needs to derive the NAF key and sends it to the client. Again, this message could be a message which only triggers the set-up of a security association, or it could contain encrypted payload data.
When the client receives the message from the NAF, it connects to the BSF transmitting the B-TID thereto, authenticates itself, and requests the remaining information necessary to derive the keying material associated with the B-TID, i.e. e.g. AUTN. After having received this information it derives the service (NAF) key and verifies the integrity of the message. As the client has to connect to the BSF, it can at the same time get all the information related to the keying material, i.e. Add-Info, key life time etc, thus reducing the amount of “administrative” information that has to be transmitted from the NAF to client.
The signalling exchange associated with this procedure, assuming the Ks generation scenario (i.e. analogous to
It may be undesirable in some circumstances to reveal the value RAND to the NAF. This may be avoided by forming the B-TID using a reference to the actual RAND value (or the effective RAND, RANDe), so that the NAF sees only the reference value. The effective RAND (RANDe) would then have to be signalled together with the AUTN from the BSF to the client. This modified procedure is illustrated in
The main advantage of the solutions described with reference to
One threat to the solutions of
A solution is to use the RAND (or derived value) in the B-TID to derive two keys Ck′ and Ik′ at the BSF. The BSF then derives a MAC key using these keys, and sends the MAC key to the NAF. This integrity key should preferably also depend on the NAF identity. Using a “fingerprint” of the other necessary information needed to derive the NAF key in the derivation of the integrity key would be one way to achieve this without having to send all the information to the UE. The NAF computes a second (short) MAC over at least a part of the data to be sent to the client, and includes the MAC in the message sent to the client. At the client, the USIM/ISIM uses the AKA algorithms to generate Ck′ and Ik′ and hence the second MAC key, and the client can then verify the message. Alternatively, the BSF can provide the keys Ck′ and Ik′ to the NAF to enable the NAF to generate the second MAC key itself. This doesn't stop replay of old message (although this could be addressed with the use of timestamps), but it does stop attackers from generating random messages.
In an alternative solution, illustrated in the signalling diagram of
According to a still further alternative solution, illustrated in the signalling diagram of
The above discussion has considered the application of the invention to the provision of service related keys to users and service nodes. Another application of the present invention relates to the provision of keys to client terminals to allow one client terminal to push messages to a peer client terminal in a secure manner, that is to say peer-to-peer (p2p) key management.
According to one solution, an initiating UE, i.e. UEA, employs the method illustrated generally in
The BSFB returns to UEA, via BSFA, a Diffie-Hellman public value for UEB, namely gNAF Key. It also returns the B-TID (containing the RAND' value used to generate the NAF Key), AUTN, and required further data. The initiating party UEA then forms a message containing its public Diffie-Hellman value, gRAND, and the information needed by the receiver to derive the KSB, the related NAF_Key, and hence the session key gRAND*NAF-Key, UEA can of course derive the same session key.
An alternative p2p key management solution is illustrated in
With this scheme, the receiving party does receive an implicit verification of the sender's claimed identity as this identity is used in the NAF_UE_Key derivation. The receiving party could also get an explicit authentication if BSFB includes a MAC based on a “NAF_Key” covering all data, as described above.
It will be appreciated by the person of skill in the art that various modifications may be made to the above described embodiments without departing from the scope of the present invention. For example, whilst the solutions presented above have been concerned with GBA, the invention has general applicability to architectures where information is to be pushed from a service provider and where the service provider and the client do not share a common secret. In another modification, where multiple solutions are implemented in parallel, the authentication request sent to the BSF contains a selector indicating which solution the NAF/UE shall employ.
This application is a Continuation of Ser. No. 13/348,343, filed Jan. 11, 2012, which is a Continuation of Ser. No. 11/305,329 filed Dec. 19, 2005, now U.S. Pat. No. 8,122,240, which is a Continuation-in-part of Ser. No. 11/248,589, filed Oct. 13, 2005, now abandoned, the entire contents of each of which are hereby incorporated by reference in this application.
| Number | Date | Country | |
|---|---|---|---|
| Parent | 13348343 | Jan 2012 | US |
| Child | 14512239 | US | |
| Parent | 11305329 | Dec 2005 | US |
| Child | 13348343 | US |
| Number | Date | Country | |
|---|---|---|---|
| Parent | 11248589 | Oct 2005 | US |
| Child | 11305329 | US |
