Hi Pankaj,
Ideally, you can intergrate the acceleration to get the velocity and get the position through double integration.
But you need to consider the errors during the integration:
1. offset
2. noise
3. bias drift due to temperature
4. tilt variation over the integration period.
...
For your case, suppose you calibrate the offset perfectly, the biggest error in your test could be the tilt variation.
You push the board by yourself, suppose there is only 1 deg tilt angle during your pushing.
It will cause the postion error as below:
0.5 * 1g * sin(1 deg) * 9.8 m/sec^2/g * 25sec ^2 = 53.4m
Please note that tilt and acceleration due to motion are indistinguishable by the accelerometer.
Hope these can help.
BR
Neil