The evolution of wireless communications is bringing into
reality the dense deployment of femto and local area cells,
which represents a challenging scenario for proving the effectiveness
of Cognitive Radio (CR) frameworks. In particular
the Dynamic Spectrum Allocation (DSA) paradigm
aims at solving the resource allocation problem in a fully autonomous
way. While in both CR and standardization lot of
effort has been spent in developing efficient DSA algorithms,
their research-oriented software radio implementation is still
disregarded. In this paper, the design approach used for the
definition of a CR-based, local area-oriented testbed platform
is introduced. In particular, the testbed design in relation to
the selected scenario and the architecture of the developed
software framework are presented. An analysis of the performance
of key software functionalities is also provided: preliminary
results show a computational load compatible with
nowadays Commercial-Off-The-Shelf computers.