A common approach is to attach strain gauges to the root of the gear's teeth. Using a proper mechanical model of the gear (e.g. FE) you can then compute other quantities of interest.

I do not know any non-contact method that could work, as inside of the housing during engagement of the gears there is usually a lot of lubricant spilling around making optical measurements difficult.

Sir, Thanks a lot for your suggestions. But are there any wireless, remote strain gauges available?

There some with slip rings that can be used on rotating shafts. I never heard of wireless models, but that doesn't mean necessarily they don't exist.

Thank you for your suggestions. I am also interested in this topic. What about using thin-film sensors? Do you have any experience with it?

The following infermation may be useful for your studies

Abstract:We propose to use a non-contact vibrational sensor to detect vibrations and deflections A gear evaluation system which includes means for measuring gear rotational speed and gear case vibrational displacement associated with gear mesh action. A signal is displacement associated with gear mesh action. A signal is developed that represents the time series average signal synchronized to the relevant gear shaft speed. This signal is passed through a bandpass filter set to pass a signal at the gear mesh frequency (for the particular gear under evaluation). After rectification the output signal provides information relating to average vibrational levels associated with gear mesh action and deviations from the average levels.

Nondestructive testing of gears

US 4872337 

www.google.com/patents/US4872337

 2) Measurement sensing devices

US 4228591 A

www.google.com/patents/US4228591

Thank you for the information. 

Thank you very much Murali Krishna sir...

Strain gauges... and if you will perform dynamic tests, you should use a telemetry system or sliding contacts...

Thank you very much Franco sir. I  was unaware of Telemetry system, will search about about its applicability..