In LTE Uplink timing is calculated at Base Station using cross correlation of PRACH when UE joins the network.
Can I use Demodulation reference signal of two successive symbols to calculate the timing slip per symbol due to combined effect of DAC at UE and ADC at Base Station miss match plus UE movement.
UE movement will add algebraically with sampling clock offset and the combined effect can be estimated along with frequency offset in frequency domain by multiplying in frequency domain Demodulation reference signal of any index with previous conjugate of demodulation of previous symbols in the same way done for DVB-T etc. Then using least square fit the slope of line will be the combined sampling clock offset. This is offset per symbol say Zeeta (Plus or minus).In say 10 seconds the total slip will accumulate and approximately 10*Zeeta..This amount can be informed to mobile using timing advance.
Since Demodulation reference signals are always present this can be done.
Since our job is to advance or slow down CP beginning we do not worry whether this contribution is from UE(Mobile) velocity or actual sampling clock offset. We need not bother about mobile position alone.
Any comments
I am not sure what is the intention of your suggestion. Are you not happy about 3GPP standard LTE system and want to try "DVB-T...etc" methods to apply them on LTE?
The LTE signal is far richer bucket of standardised signals and algorithms so why give it up.
I am not suggesting any modification of LTE.Since the receiver design is up to individual and anyhow Demodulation reference signal for each uplink UE is available in different frequency band I am suggesting it to use in frequency domain to calculate sampling clock offset (Combined effect of ADC/DAC slip and vehicle movement) .This will give give slip in time for one symbol. We can calculate by multiplying any time the total adjustment needed.This is same as joint frequency offset/sampling clock offset estimation used in DVB-T (IEEE by Meyr).
But I found for maximum velocity of 30m/sec by UE in 10 seconds 300 metre distance will be covered.This will cause a delay of 1 micro sec for electro magnetic waves.So the rate is 1 micro sec/10 sec=0.1 ppm.
But I am doubtful whether this low ppm can be estimated correctly by the method.
This idea I got some where from under water communications.But there velocity of sound will be low.So the slip will be much larger than 0.1 ppm.
I do not know in LTE with out using Sounding signal which is optional how the timing update can be done to track continuous movement of UEs.
If we try to measure the phase shift and time delay we should know the symbol number which is not needed in my method.
I don't have direct experience with DM-RS but I know LTE uses it to track each UE timing offset and frequency offset and I assume that includes tracking a moving UE case since these signals are always received.
SRS is for a different purpose, specifically it is a check on all bandwidth (not just one UE) so that best bandwidth section is allocated for a given UE assuming such choice is available and yes it is optional.
This is one link that explains use of DM-RS for timing/frequency offset: