Dear all, I am trying to implement a time-synchronization algorithm on Telosb motes using TinyOS 1.x. I am using ClockTimeStamping to timestamp the messages at the MAC layer, as well as LocalTimeMicro to have a micro-second resolution. The problem is that when I run the test I get the synchronization error divided into groups (usually 3) that are spaced from each other by about 31us. Here are some of the readings I got: 3.0 -1.0 2.0 -1.0 35.0 31.0 34.0 31.0 2.0 31.0 -32.0 2.0 -3.0 29.0 -31.0 -3.0 This behavior can be clearly seen by plotting a histogram. The histogram I got (for 1000 readings of 1 run) shows that the synchronization error has 3 groups that are centered at 0us, 31us, and -32us with Gaussian distribution around each center ranging +-8us and the rest is empty spaces between the groups. 1- Does the MSP430 use the 32kHz TimerB to control its operations? Because if so, then this means that the transmission or reception request by the CC2420 is delayed (ex: by an interrupt) until the next clock tick which is 30.5us later and then served (while the timestamp is already written).So, is there a way to work around that? 2- Did anybody happen to face such an issue? Can you please guide me if there is a solution? I have tried my code on Mica2 (after changing the time components) and it worked just fine. I have read some posts about TinyOS 2.x being provided with new codes for time-stamping (PacketTimeStamping) that do not seem to have such behavior , so if there is no solution is it recommended to move to TinyOS 2.x, or is it still a poorly explored land? Kind regards, Hussein _______________________________________________ Tinyos-help mailing list Tinyos-help@millennium.berkeley.edu https://www.millennium.berkeley.edu/cgi-bin/mailman/listinfo/tinyos-help
