Re: [MEXT] Well-known problem with authentication/etc. in wirelessnetworks

[<[e-mail]>]

On 8/26/11 3:11 PM, "ext Julien Laganier" <[e-mail]> wrote:

>Hi Raj,
>
>On Fri, Aug 26, 2011 at 1:09 PM,  <[e-mail]> wrote:
>>
>> Hi Julien,
>>
>> On 8/26/11 3:00 PM, "ext Julien Laganier" <[e-mail]> wrote:
>>
>>>Hi Raj,
>>>
>>>On Fri, Aug 26, 2011 at 12:01 PM,  <[e-mail]> wrote:
>>>>
>>>> Hi Charlie,
>>>>
>>>> On 8/26/11 1:38 PM, "ext Charles E. Perkins" <[e-mail]>
>>>>wrote:
>>>>>
>>>>>>> Thus I am still not sure what the problem is.
>>>>>
>>>>>
>>>>>The problem is that they can't do very effective handovers.
>>>>>Worse, they are designing _per-application_ handover systems.
>>>>>This is wrong by most reasonable engineering standards,
>>>>>regardless on the positive effect it might have for
>>>>>standards junkies and permanent employment for engineers.
>>>>
>>>> Effective handovers between what networks? Handovers within the scope
>>>>of
>>>> an HSPA or LTE access for example work fine.
>>>> If you are referring to handovers between 3G accesses and wifi
>>>>(non-3GPP
>>>> access) then yes.
>>>> But the handover performance in such a scenario is hampered by other
>>>> factors such as latency in connectivity and authentication etc.
>>>
>>>In the latter (handover between 3GPP and non-3GPP), given that the
>>>source and target system are accessed by different radio systems, I do
>>>not see handover performance has a factor hampering the usability or
>>>desirability of the inter-system handover scheme in use.
>>
>> You mean the impact (in terms of lower performance) of handovers across
>> 3GPP and non-3GPP handovers does not really matter because those
>> applications that rely on high performing handovers will anyway not rely
>> on such?
>
>No I meant that because the MN has different radio systems, it can
>turn one on and send a BU while still receiving/sending data via the
>old one, and thus MIP handover performance isn't in a critical path in
>that situation. For single radio the situation is obviously different.

Agree. Such a scenario would essentially result in a make-before-break
handover and hence the latency resulting from attach and authentication
would not be as much of a concern. Thx for the clarification.

>
>--julien