On the Feasibility of Instantaneous Power Adaptation Under Incomplete CSI for Secure Wireless Communication

Abstract

In secrecy-constrained wireless systems with cochannel interference, artificial noise (AN)-aided transmission relies on power allocation between information and noise signals to support reliable communication while accounting for secrecy constraints. While instantaneous power adaptation can exploit channel variations, its applicability is fundamentally limited by incomplete channel state information (CSI), particularly the lack of instantaneous knowledge of interference and passive eavesdropper channels. This raises a critical question: when is instantaneous power adaptation under incomplete CSI actually feasible and beneficial? To address this, a mixed-CSI-based power allocation framework is developed, combining instantaneous knowledge of the legitimate channel with statistical characterization of the unknown links. By introducing a statistically designed surrogate, an implementable instantaneous power allocation rule is obtained. The resulting performance is governed by the feasibility probability, which quantifies how often the induced power allocation satisfies the reliability constraint while accounting for secrecy through statistical design. The proposed scheme thus behaves as a probabilistic realization of the full-CSI system, approaching its performance as this probability increases. A key structural insight is that, under incomplete CSI, the reliability condition is enforced per realization, whereas secrecy is captured only statistically, leading to instances where reliability is satisfied while secrecy is not. These findings establish that instantaneous adaptation under incomplete CSI is not inherently beneficial, but is dictated by the feasibility of the induced decisions. Simulation results validate the analysis.