IN order to improve performance of the original normalized subband adaptive filter algorithm with the step-size scaler (SSS-NSAF) when identifying sparse impulsive response, the proportionate SSS-NSAF (SSS-PNSAF) algorithm and improved proportionate SSS-NSAF (SSS-IPNSAF) algorithms are given by utilizing common proportionate strategy. Even though the performance of the SSS-PNSAF algorithm is improved in sparse system, its convergence rate even slower than the original SSS-NSAF algorithm when the impulse response is disperse. For possessing great performance of the SSS-PNSAF algorithm in sparse impulse response and retaining merit of the SSS-NSAF algorithm in dispersive impulse response, the gain-combined proportionate SSS-NSAF (GC-SSS-PNSAF) algorithm is proposed by combining weight coefficient vectors of these two algorithms with a variable mixing parameter. The mixing parameter is indirectly updated through a modified sigmoidal activation function by using stochastic gradient method which minimizes the power of the system output errors. Furthermore, variable control parameter (VCP) mechanism is introduced to the GC-SSS-PNSAF algorithm to overcome the trade-off issue between fast convergence rate and low steady-state error. Numerous simulation experiments confirm the superiority of these proposed algorithms.