The design of the individual components of proposed methods will be evaluated both using extensive simulations and via testbed evaluations. For the testbed evaluations, the CR network testbed will be designed and implemented in our labs. The structure of the CR network testbed is illustrated in Figure 7, which consists of CR users that are implemented through the cognitive transceiver unit (CTU), CR base station, and various primary network models.

The cognitive capability required for CR networks will be emulated using a cognitive transceiver unit (CTU). All of the functional blocks in the CTU, including MODEM, RF/Analog front end, and proposed cross-layer CR functionalities are designed to be operated by software (S/W) and/or field programmable gate array (FPGA) format for acceleration of response time, as well as for ease of management and expansion. This component consists of an RF front-end, spectrum sensing processor and an FPGA board, which provides software interface for the development spectrum management functionalities. The CTU includes functional spectrum sensing technology, and is capable of instant testing/evaluation of the algorithm level of the communication system and of the RF/Analog System/IC. Also, the CTU will have a network operation capability to validate CR based user scenario models. Once a core IC or new algorithm has been developed, they will be incorporated into the FPGA board. Consequently, the proposed techniques will be implemented on the CTU for comprehensive evaluations.



Figure 7. Testbed for cognitive radio networks.

The primary network activity that is necessary for accurate evaluation of the cognitive radio network will be provided through three different techniques. First of all, the Georgia Tech WLAN that is operating in our working environment will be regarded as a primary network for the CR nodes. Moreover, the existing wireless mesh network testbed available at our labs will perform as a secondary primary network at the 2.4 GHz range. As a result, two simultaneously operating primary networks will be emulated at the same spectrum band. Finally, the RF signal generator, which can generate various types of standard signals including VHF/UHF TV signals, CDMA, GSM, Wi-Fi and WiMAX, will be used to emulate existing large scale networks in the 54 MHz - 6 GHz range. Using a combination of these settings, we will perform thorough experiments to characterize the performance of the proposed communication protocols by deploying the testbed indoor as well as outdoor.

As shown in Figure 7, at the 2.4 GHz range, 15 non-overlapping channels can be used (3 for IEEE 802.11g, and 12 for IEEE 802.11a). Also, CTU will have the capability to use all other available channels between 54MHz and 6GHz. Hence, we will evaluate our spectrum management solutions and the communication protocols using these channels. These channels constitute a large subset of the available channels foreseen for cognitive radio networks and they accurately represent intrinsic challenges of the cognitive radio networks.

The CTU continuously monitors the available channels in the CR network testbed. The CTU will also be implemented with an interface to the CR nodes emulating the cognitive capabilities of the cognitive radio physical layer and providing spectrum information to the nodes. Also, the testbed will support network operation capability to validate CR based user scenario models such as primary network access, CR ad hoc and network access.

The testbed will serve, not only as a technical development platform, but also as an educational tool that provides insight and deep understanding of cognitive radio technology. We believe this testbed is a world first for the study of a full S/W overlay cross-layer system platform that can deal with wireless communication, digital multimedia signals up to protocol level, as well as incorporating System-on-Chip prototypes developed in-house.