The focus of this work is put on a cognitive network wherein the secondary users (SUs), based on bit-interleaved-coded orthogonal frequency division multiplexing, share the same band of primary users (PUs). In this context, we develop a distributed power allocation (PA) strategy which minimizes the coded packet error rate (PER) of the noncooperative SU links under the constraint of satisfying the interference limit at PU receivers. The proposed PA problem is formulated as a noncooperative game, relaxed by applying a pricing mechanism and then solved by an iterative algorithm. On top of the PA strategy, an adaptive modulation and coding scheme is also applied to further boost performance. Numerical simulations show the effectiveness of the proposed algorithms when tested in typical wireless scenarios.