Cognitive Infocommunications (CogInfoCom) studies the connections between infocommunications and cognitive experiences. This paper presents a practical performance evaluation of an active antenna system (AAS) based on tests performed in a field trial environment with a cellular high-speed packet access (HSPA) network. Cellular networks face the challenges of constantly growing data traffic and demands to reduce network costs, which call for new approaches for optimizing network performance. The demands in different parts of the cellular network change because of the spatially and temporarily varying usage characteristics. An AAS can offer a solution to these demands with flexible beamforming to respond to the changing situations in the network. This study shows that the AAS can gain improvements in system performance in terms of achievable throughputs. The spatial cell isolation needs to be taken care of in order to maintain optimal system operation by avoiding inter-cell interference. Since the changing radio network requires continuous optimization, an AAS needs intelligent control for full advantage. This can be accomplished with cognitive decision making by exploiting Self-Organizing Networks (SON) features. Later, network performance optimization with an AAS could use "cognitive engines" to intelligently add sectors where capacity is needed and switch them off when the capacity demand decreases. Cognition combined with an AAS could optimize the network and adjust the AAS parameters according to different needs in the changing environment.