[net2-wg] dhv release timeline

[dangtx at pdx.edu]

Quoting "Razvan Musaloiu-E." <razvanm at cs.jhu.edu>:

> Hi!
>
> On Sun, 23 Aug 2009, dangtx at pdx.edu wrote:
>
>> There one assumption that DIP doesn't explicitly assume that is all  
>> the nodes have the same number of keys. DHV requires that  
>> assumption in order to work. However, both DIP and DHV don't yet  
>> explicitly address the case where nodes have different number of  
>> keys/programs, which can be part of the heterogeneity umbrella.
>
> I think the assumption that all the nodes have the same keys is  
> reasonable. All the nodes need to participate in dissemination of  
> all the keys anyway. If the nodes only propagate the keys they care  
> about then some updates might never reach some nodes.
>
> Quick q: what is a program? :P

The right word should be item, which can be any piece of data you want  
to dissemination.

>
>> About the two questions:
>>
>> + The size of the data for the test is 2 bytes. As long as the data  
>> can fit into one tinyos message, I think the results won't be much  
>> different.
>
> It would be good to test with some bigger data because when the data  
> is small a few big Drip would also work pretty well probably. :-)
>

This raises an interesting question about the scope of dissemination  
protocol. Should segmentation is part of a dissemination protocol?  
I'll post a separate thread on this.

What does "a few big Drip" mean?


>> + keys to be updated are picked randomly at compile time.
>>
>> I will work with Kaisen to write up an user guide and update the  
>> tutorial. If there's no problems, I think TEP 118 should also be  
>> updated.
>>
>
> What updates does the TEP 118 need? That TEP is final so the only  
> thing can happen with it is to deprecated. A new one would need to  
> be created if changes has to be made (we are facing the same problem  
> with TEP 105, the one for LPL :P).
>

I think we should update the implementation section to include dhv.  
I'm not sure about the process for TEP though.


Thanh


> All the best!
> Razvan ME
>
>> Quoting "Razvan Musaloiu-E." <razvanm at cs.jhu.edu>:
>>
>>> Hi!
>>>
>>> I just briefly looked the results but it seems that DHV is a  
>>> better than DIP all the time! :-)
>>>
>>> Two questions: what is the size of a data associated with the key?  
>>> When an update is done the new value is picked randomly?
>>>
>>> --
>>> Razvan ME
>>>
>>> On Sat, 22 Aug 2009, Omprakash Gnawali wrote:
>>>
>>>> This is great. Can I ask you and Kaisen to work together on writing a
>>>> short document on how users should decide which protocol
>>>> (drip,dip,dhv) to use? Research-type explanation is ok, but it is
>>>> better to write an explanation for someone who wants to pick a
>>>> dissemination protocol for a particular application.
>>>>
>>>> You should also update this tutorial:
>>>> http://docs.tinyos.net/index.php/Dissemination
>>>>
>>>> - om_p
>>>>
>>>> On Sat, Aug 22, 2009 at 2:00 AM, <dangtx at pdx.edu> wrote:
>>>>> I have done the following cvs commit
>>>>>
>>>>> + add tos/lib/net/dhv
>>>>> + add apps/tests/TestDhv
>>>>> + update appds/tests/TestDissemination/Makefile with flag for dhv
>>>>>
>>>>> Also please refer to
>>>>> http://web.cecs.pdx.edu/~dangtx/pub/ewsn09/DIPnDHV.pdf for the
>>>>> protocol tests.
>>>>>
>>>>>
>>>>> Thanh
>>>>>
>>>>>
>>>>>
>>>>>
>>>>
>>>
>>>
>>
>>
>>
>
>