[Owasp-singapore] [security-77] IBM Researcher Solves Longstanding Cryptographic Challenge,Discovers Method to Fully Process Encrypted Data Without Knowing its Content; Could Greatly Further Data Privacy and Strengthen Cloud Computing Security

Wong Onn Chee ocwong at usa.net
Tue Jun 30 21:36:20 EDT 2009


Thanks for the clarification, Andrew.

To clarify further, for Pt (b), I am referring to business process 
triggers, such as triggering inventory replenishment thresholds and etc, 
not actually IS triggers.


On 07/01/2009 07:25 AM, andrew wrote:
> bruce should be able to explain it better and it had been a
> cryptologist's problem for many years. most encryption scheme suffer
> from data corruption when calculations are imposed on them.
>
> yes, this is an oversimplistic example. just a little background:
>
> for a long time, we have homomorphic encryption schemes, but never
> fully homomorphic ones. In order to be fully homomorphic, you must be
> able to add and multiply.
>
> the problem with existing schemes for example, RSA is that it is
> homomorphic but nof fully, meaning it can only use multiplication. the
> other problem is that it has deep circuits.
>
> so to be practically used in the real world without data corruption,
> esp after several applications, you need to find an encryption scheme
> which is self evaluating. So you need to refresh it after the error
> vector becomes too great. So you need to decrypt it to do more
> calculations. This creates another puzzle which mathematicians
> struggled for years.  You need the secret key to do so. So instead he
> homomorphically decrypt it. you need some proofing here to explain how
> it works. since most of us are not mathematicians / cryptographers, i
> shall skip that (you can read it from his papers). in short, this is
> where he uses a decryption key to compare with the function without
> one with acceptable rounding so it won't go off.
>
> next he needs to find a known attack for it. because the general
> perception is that x(enc)/y(dec) is not a large enough value. so he
> compares it with existing schemes like RSA or ElGama.
>
> So there's a lot of thought process to come up with this sort of
> scheme. He did it differently because instead of just deriving a
> function, he changed the process to make it easier, if that makes
> sense.
>
> so in short, your concern for a) have been addressed.
>
>
> b) From the perspective of cloud computing, you may be right. But this
> is subject to implementation. For example, a SaaS IDS service with an
> expert system,
>
> a remote attack that allows admin access = 50 pts
> a 90% false negative attack = 10 pts
> a current valid vulnerability = 30 pts
> a critical asset being attacked = 50 pts
>
> again, this is oversimplistic and i know fundamentally flawed from a
> product design perspective (which is not the basis of our discussion).
>
> what the appliance do is to send all the encrypted values or encrypted
> IDs of each property to be calculated. The expert system adds,
> multiply to deduce the criticality of the attack and sends back the
> encrypted information of the attack to the client end appliance which
> decrypts it. Do take note that the encryption scheme E fulfills the
> criteria of not getting corrupted after several rounds of
> calculations.
>
> there's no problem with y being used b'cos it is presented as x only
> on a need to know basis. of course again in this example, the analyst
> at the provider end will not be able to analyse the actual details of
> attack, but there is always a workaround. The customer can expose x to
> the provider without encryption.
>
>
>
> On Wed, Jul 1, 2009 at 1:33 AM, Wong Onn Chee<ocwong at usa.net>  wrote:
>    
>> Hi Andrew,
>>
>> Thanks for the layman explanation. Appreciate it a lot.
>> Good for ignorants like me. :-)
>>
>> I still see some challenges with this approach, so feel free to shot them
>> down if you disagree:
>>
>> a) I understand y*30% is a simplified example.
>> However, most cloud apps are much more complex than the above example.
>> Hence, to obtain the entire chain of calculations may be a problem.
>> In addition, there is also the question whether the cloud app provider will
>> expose their own algorithms to their customers.
>>
>>
>> b) Assuming the output of the calculations (in this case, the result of
>> y*30%) is measured against some thresholds.
>> If the output exceeds the threshold, further actions may be triggered.
>>
>> However, if x*30% is within the threshold, but y*30% exceeds the threshold,
>> won't that result in false triggers or unnecessary further actions? And the
>> reverse may also occur.
>>
>> Imagine that the business mgt missed out critical business intelligence
>> information becos y is used, instead of the real x?
>> Won't the reliability of cloud computing be questioned?
>>
>> I have brought this up to in my discussion with you-know-who in MHA about
>> this gap in cloud computing.
>> Using encryption makes sense only if you use the cloud to host emails and
>> store files, but that is not cloud computing to me.
>> Existing services, such co-location, dedicated servers, managed services,
>> can already address the needs of email hosting and file archival.
>> Cloud computing is about sharing the compute load, not just for storage or
>> email purposes.
>>
>> Also, it does not make sense to ask for dedicated machines from the cloud to
>> avoid this encryption loophole, as the basis of cloud is about sharing of
>> resources.
>> Dedicated machines from the cloud is no different from dedicated hosting.
>> As Shakespeare said, a rose smells as nice by any other name.
>> So dedicated machines from cloud is equal to dedicated hosting to me. ;-)
>>
>> Just my 2-cent worth on this topic.
>> Pardon my non-mathematical view of the problem.
>> Me more realist lah. :-)
>>
>>
>>
>>
>> On 06/30/2009 01:25 PM, andrew wrote:
>>
>> i just realize i didn't explain a little about the crypto as i
>> understand it so far (without digesting the paper yet):
>>
>> I have a function
>>
>> f(x) = y
>>
>> I send y to my accountant
>>
>> accountant does tax for me
>>
>> y * 30%
>>
>> accountant has no knowledge of x, which is my actual pay
>>
>> so I do a
>>
>> f(y*30%)-1 = x*30%
>>
>> if you are familiar with RSA, you know that the 3 boys were finding a
>> function that fulfills certain criteria using large primes which are
>> infinitely available and difficult to factorise which is a one way
>> function unless a trapdoor is provided. So what this function has to
>> fulfill is to match all required mathematical function while
>> encrypting to the decrypted form and fulfilling encryption
>> requirements at the same time. Hope that makes sense.
>>
>>
>> -1 in that formula shd b superscript which means a reverse function.
>> sorry i'm lazy to clean up but that pretty much explains it.
>>
>> P.S. I gotta do my tax tomorrow that's why the example.
>>
>> On Tue, Jun 30, 2009 at 2:15 PM, andrew<quickt at gmail.com>  wrote:
>>
>>
>> A more detailed article:
>> http://www.forbes.com/forbes/2009/0713/breakthroughs-privacy-super-secret-encryption.html
>>
>> His paper:
>>
>> http://delivery.acm.org/10.1145/1540000/1536440/p169-gentry.pdf?key1=1536440&key2=8197236421&coll=GUIDE&dl=&CFID=37558608&CFTOKEN=60907897
>>
>> Still trying to digest it.
>>
>>
>> On Tue, Jun 30, 2009 at 12:27 PM, Johnny Wong<johnnywkm at gmail.com>  wrote:
>>
>>
>> Maybe can send a note to Bruce Schneier and ask him for his assessment.
>>
>> At 09:41 AM 30-06-09, Wong Onn Chee wrote:
>>
>>
>> Anyone who can shed some more light on this?
>>
>> http://www-03.ibm.com/press/us/en/pressrelease/27840.wss
>>
>> Sounds like the missing link for cloud security. :-)
>>
>>
>>
>>
>>
>> --
>> Please Note: If you hit "REPLY", your message will be sent to everyone on
>> this mailing list (security-77 at meetup.com)
>> http://security.meetup.com/77/
>> This message was sent by Wong Onn Chee (ocwong at usa.net) from The Singapore
>> Security Meetup Group.
>> To learn more about Wong Onn Chee, visit his/her member profile:
>> http://security.meetup.com/77/members/1756147/
>> To unsubscribe or to update your mailing list settings, click here:
>> http://www.meetup.com/account/comm/
>> Meetup Support: support at meetup.com
>> 632 Broadway, New York, NY 10012 USA
>>
>>
>>
>> --
>> Please Note: If you hit "REPLY", your message will be sent to everyone on
>> this mailing list (security-77 at meetup.com)
>> http://security.meetup.com/77/
>> This message was sent by Johnny Wong (johnnywkm at gmail.com) from The
>> Singapore Security Meetup Group.
>> To learn more about Johnny Wong, visit his/her member profile:
>> http://security.meetup.com/77/members/5695170/
>> To unsubscribe or to update your mailing list settings, click here:
>> http://www.meetup.com/account/comm/
>> Meetup Support: support at meetup.com
>> 632 Broadway, New York, NY 10012 USA
>>
>>
>>
>>
>> --
>> Please Note: If you hit "REPLY", your message will be sent to everyone on
>> this mailing list (security-77 at meetup.com)
>> http://security.meetup.com/77/
>> This message was sent by andrew (quickt at gmail.com) from The Singapore
>> Security Meetup Group.
>> To learn more about andrew, visit his/her member profile:
>> http://security.meetup.com/77/members/8358574/
>> To unsubscribe or to update your mailing list settings, click here:
>> http://www.meetup.com/account/comm/
>> Meetup Support: support at meetup.com
>> 632 Broadway, New York, NY 10012 USA
>>
>>
>>
>>
>>
>> --
>> Please Note: If you hit "REPLY", your message will be sent to everyone on
>> this mailing list (security-77 at meetup.com)
>> http://security.meetup.com/77/
>> This message was sent by andrew (quickt at gmail.com) from The Singapore
>> Security Meetup Group.
>> To learn more about andrew, visit his/her member profile:
>> http://security.meetup.com/77/members/8358574/
>> To unsubscribe or to update your mailing list settings, click here:
>> http://www.meetup.com/account/comm/
>> Meetup Support: support at meetup.com
>> 632 Broadway, New York, NY 10012 USA
>>
>>
>>
>>
>>
>>
>>
>> --
>> Please Note: If you hit "REPLY", your message will be sent to everyone on
>> this mailing list (security-77 at meetup.com)
>> This message was sent by Wong Onn Chee (ocwong at usa.net) from The Singapore
>> Security Meetup Group.
>> To learn more about Wong Onn Chee, visit his/her member profile
>> To unsubscribe or to update your mailing list settings, click here
>>
>> Meetup Support: support at meetup.com
>> 632 Broadway, New York, NY 10012 USA
>>      
>
>
>
> --
> Please Note: If you hit "REPLY", your message will be sent to everyone on this mailing list (security-77 at meetup.com)
> http://security.meetup.com/77/
> This message was sent by andrew (quickt at gmail.com) from The Singapore Security Meetup Group.
> To learn more about andrew, visit his/her member profile: http://security.meetup.com/77/members/8358574/
> To unsubscribe or to update your mailing list settings, click here: http://www.meetup.com/account/comm/
> Meetup Support: support at meetup.com
> 632 Broadway, New York, NY 10012 USA
>
>
>
>    
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://lists.owasp.org/mailman/private/owasp-singapore/attachments/20090701/a3628475/attachment.html 


More information about the Owasp-singapore mailing list