[Tinyos-help] About the performance evaluation of CTP

[Philip Levis]

On May 7, 2008, at 5:00 AM, Xiaojun Zhu wrote:
> Hi everyone,
>          Has any work been done towards the performance evaluation  
> of CTP in tinyos2.x?I failed to find one. I mean a comparative stduy  
> of CTP vs. other protocols.
>

The only published results on CTP are in HotNets 2007 ("Four Bit  
Wireless Link Estimation"). These results report CTP's datapath  
efficiency versus MultihopLQI and MintRoute. There are results on  
CTP's datapath and control path efficiency, but these are in a paper  
under submission. The short answer is that CTP can cut lost packets by  
90% and improve efficiency (transmissions per packet delivered at the  
root) by 20-30%, while cutting control traffic by two thirds. It does  
this while remaining radio chip independent. The tradeoff is that it  
is more complex (in terms of code size and RAM) than MultihopLQI. Once  
we've nailed all of the last issues in CTP, I think the plan is to do  
a clean re-implementation: right now, it's code which has been  
iterated on for two years, so is a bit more convoluted than it needs  
to be.

Phil