Adaptiveness of femto base station's (FBS's) transmission power (TP) is crucial in determining Quality of Service and ecofriendliness of a network. Joint power optimization and admission control problems have been reformulated to identify the TP of v already deployed FBSs, and the problem is shown to be computationally hard. Accordingly, an adaptive distributed heuristic called power search algorithm (PSA) is proposed. A learning algorithm analytically identifies the received signal strength‐based coverage of each FBS in all directions during planning. A 3‐dimensional reference matrix, ie, REF, for each FBS, is thereby formulated and stored in the corresponding FBSs. Power search algorithm handles call admission/termination at run time. For call admission, PSA identifies the serving FBS and the required minimum TP. A suitable data structure is also maintained by PSA for efficient call handling. A new call is dropped if its admission degrades the Quality of Service of existing end‐users. For call termination, change in TP is triggered if the highest power level of the FBS was being exercised to serve this particular user. The worst‐case run‐time complexity of PSA turns out to be O(log2N), where N is the number of TP levels of a particular FBS. Comparing with existing heuristics, improved performance of PSA includes highest throughput and signal‐to‐interference‐plus‐noise ratio while minimizing starvation and cumulative TP with lower complexity even for high end‐user density. Exhaustive simulation reveals that the probability of an end‐user having signal‐to‐interference‐plus‐noise ratio above 0 dB is 0.83 even in the worst case scenario. Accordingly, the proposed PSA is claimed to be superior.