I previously posted "a clock synchronisation problem" but now think that I was asking the wrong question. My aim is to look for small errors in time of the local computer compared to a more precise time server over a period of tens of minutes.
The process of getting time from a remote computer has several steps where time stamps are collected from the remote computer and the local computer;
t0 = Originate Timestamp—The local time at which the request departed from the client for the service host.
t1 = Receive Timestamp—The server time at which the request arrived at the service host.
t2 = Transmit Timestamp—The server time at which the reply departed from the service host for the client.
t3 = the client's timestamp of the response packet reception.
The formula for the time correction ($\theta$) from Wikipedia is (1) Wikipedia NTP or (2) given to me by Ross Millikin which he says makes it clear that it is the difference between the two clocks at the midpoint of the exchange.
(1)$$ \theta = \frac{(t_1-t_0)+(t_3-t_2)}{2} $$
(2)
$$\theta = \frac {t_2+t_1}2-\frac {t_3+t_0}2$$
Am I right in thinking that this is just the difference between the average of the server and client, send and receive time stamps? But as the transmit time might be different, as shown in my figure below, I think that the averaging will obscure the data I am trying to acquire? Also, the interval between t2 and t3 would be useful but as they are recorded on separate computers I think that the interval between them can not be known except that there is a lower limit near zero and an upper limit, which is the round trip delay, given by (t3-t0)-(t2-t1)?
