[net2-wg] DHV / DIP

[dangtx at pdx.edu]

Short answer: There was a mistake in our measurements and a bug in DHV  
in the paper.

Long answer:
  * The measurements: What we did was compiling the TestDip and  
TestDhv program with 0 total number of keys (T) and 0 number of new  
keys (N) and thought that it would give us the baseline for ROM, RAM,  
and code size.  We realized that these are useless measurements  
because we want to observe it with different number of keys.

  * The bug in DHV: In the code, there is a place to set the number of  
tinyos messages  to an uint8_t using a formula [(T-1)/max_length] + 1.  
When T is set to 0, this formula sees (T -1)=-1 as a large integer and  
truncate to unsigned type and causing the increase in RAM usages.

The current DHV does not have this bug.

Thanh


Quoting Philip Levis <pal at cs.stanford.edu>:

> The paper suggested that the RAM size for DHV grows significantly:  
> can you explain the difference between the results here and those in  
> the paper?
>
> Phil
>
> On Aug 11, 2009, at 11:56 PM, dangtx at pdx.edu wrote:
>
>> <ProtocolComparison.pdf>
>
>
>