A resource allocation method and a base station which includes: acquiring, by the base station, channel state information (CSI) of a current scheduling timeslot; obtaining a compensation factor according to the CSI and a preset packet error ratio, where the compensation factor is a value corresponding to the preset packet error ratio in an inverse cumulative distribution function of a non-central chi-square random variable and a non-central parameter; replacing the CSI with the compensation factor, and determining a functional relationship between rate and power of sending data by the base station if a packet error ratio of the base station is less than or equal to the preset packet error ratio; and determining, according to the functional relationship and a generalized dynamic back pressure algorithm, the rate and total power of sending data by the base station, so as to minimize the total power of the base station.