List Info

Thread: comments on drafts using CGA
comments on drafts using CGA
user name
 2006-10-22 12:22:53 

  


  

    
Thanks for your comments, it helps me a lot. Please see my comments inline.


 


Thanks,


Zhen

 


On 10/22/06, James Kempf <docomolabs-usa.com" target="_blank">
kempfdocomolabs-usa.com&gt; wrote: 

Zhen,

A few years ago, some of us published some drafts on how to use ID crypto
for address security. Basically it came down to using the address as the 

public key. The problem is the need for provisioning of the private key by
the KDC. The two problems with this were 1) the KDC knows the private key
and therefore could compromise it and 2) the KDC and the client need to 

support some other kind of cryptosystem to provide confidentiality during
the key distribution. The first problem can be limited using a couple of
techniques, like HIDE (Hierarchical Identity-based Encryption) but the 



 


For the first problem: the KDC is much more robust than client with respect to key compromise. Yes, if the KDC is compromised, all the keys are disclosed. But I think the KDC is the infrastructure of the network, such as Certificate Authority and AAA. We rely on those infrastructures so the assumption has always been that 'the infrastructure is secure enough'. If they are assumed compromised, any mechanisms MUST fail. 




second one seems insoluble. Given that RSA is a logical choice for the
cryptosystem between the KDC and the client, why not simply use CGA in the 

first place? From a system perspective, ID crypto is like Kerberos, and it
suffers from the same limitations when used for deployment scenerios that
are open-ended, i.e. not limited to particular organizations or other 

groups. Possibly you have some new insights into solving these problems, I
have not read your draft so I cannot say.


&nbsp;


Simply using CGA in the first place is not a problem. But what if we have more efficient and effective substitutions. We try to promote what CGA has guaranteed.


 


As for the open-ended deployment scenarios, we have not considered yet.  Do you have suggested reference?



Regarding your other criticisms of the drafts that use CGAs, I haven't
looked at the drafts with this perspective in mind, but if they are not 

abiding by the guidelines in RFC 3872, then they probably need some work.


&nbsp;


You meant RFC 3972? Yes, I think we should have more considerations rather than just assert that we will take advantage of CGA. 



  

  
  
   
     
    






 [1]











   
 





about | contact  Other archives ( Real Estate discussion Medical topics )