In the recent researches, some methods have been proposed for interpolating the diffusive field values along space, but they either do not use point-wise sampling or deploy large number of sensors in space to keep themselves near Niquist’s sampling rate. In this research we propose some algorithms in which we benefit from temporal-spatial correlation of field values and try to reduce the number of sensors deployed along space. First we will consider the extreme case with only one sensor in space and use the whole time line to estimate the signals and second the discrete time for which we use M sensors in space and K time samples to reconstruct the field. Finally the stability of the proposed methods will be explored and the best will be chosen in the sense of being the most stable.