Re: [Rucus] Any further comments on draft-york-sipping-spit-similarity-scenarios-00.txt ?

[Hannes Tschofenig <[e-mail]>]

I agree with you, Saverio. Nevertheless it is a good idea to keep the 
scenarios Dan has described in his document in mind and to consider them 
in the overall solution.


Saverio Niccolini wrote:
> Dear Hannes,
>
>   
>> I like your document, as mentioned previously already. I 
>> shows some of the limitations with systems that use learning 
>> of past behavior as their basis for future actions. Some of 
>> the statistical techniques belong to this category.
>>     
>
> we all agree statistical techniques do have limitations, but
> they are also powerful and this does not mean they should not
> be considered in the RUCUS work
>
> I remember all the list of methods indicated by Henning during
> the BOF as most promising (check his ppt slides):
> -- identity-based
> -- statistics
> -- price-based
>
> For some of the scenarios Dan describes in his draft the statistical
> techniques will never come into play:
> -- "emergency notifications": if you have strong identities in place
> and you are sure nobody can steal them, you just need to insert the
> URIs that are allowed to send "emergency notification" into the
> white list --> no need to apply statistics, just work on policy and
> identity-based mechanisms (statistics are useful here to check misbehaviours)
>
> -- Urgent Notification Systems: a bit more complicated but there is always
> a previous relationship between caller and callee, thus also here
> one could get around them without statistics but with policies and identity-based
> mechanisms
>
> -- for the other systems I would always repeat what I just wrote:
> apply policies and identity-based mechanisms
>
> The statistics are good when ther eis no previous relationship between
> the caller and the callee and this is why they are useful
>
> Thus if you want to choose the three class of mechanisms that RUCUS should
> suggest just choose these three..
>
> Cheers,
> Saverio
>
>   
>> However, the most important question is: What should happen 
>> with the document at the end?
>> Furthermore, the document points to potential problems to a 
>> specific category of solutions. For systems that do not rely 
>> on learning techniques these scenarios are not a problem as such.
>>
>> Ciao
>> Hannes
>>
>>
>>
>> Dan York wrote:
>>     
>>> SIPPING & RUCUS,
>>>
>>> Because I've had some various pieces of feedback on the 
>>>       
>> document, I'm 
>>     
>>> going to rev draft-york-sipping-spit-similarity-scenarios 
>>>       
>> in the next 
>>     
>>> week or so to incorporate that feedback.  Since I'm doing so, I 
>>> thought I'd just ask.... anyone else have comments on:
>>>
>>>    
>>>
>>>       
>> http://tools.ietf.org/id/draft-york-sipping-spit-similarity-scenarios-
>>     
>>> 00.txt
>>>
>>> Additional scenarios you think I should include.... points of 
>>> clarification... whatever?
>>>
>>> Thanks,
>>> Dan
>>>   
>>>       
>> _______________________________________________
>> Rucus mailing list
>> [e-mail]
>> https://www.ietf.org/mailman/listinfo/rucus
>>
>>     
>
>
> ============================================================
> Dr. Saverio Niccolini
> NEC Laboratories Europe, Network Research Division	
> Kurfuerstenanlage 36, D-69115 Heidelberg
> Tel.     [phone]
> Fax:     [phone]
> e-mail:  [e-mail] <-- !!! NEW ADDRESS !!!
> ============================================================
> NEC Europe Limited Registered Office: NEC House, 1 Victoria
> Road, London W3 6BL Registered in England 2832014
>  
>   

_______________________________________________
Rucus mailing list
[e-mail]
https://www.ietf.org/mailman/listinfo/rucus