Cognitive GSM OpenBTS Yuva Kumar S., Meghan S. Saitwal, Mohammed Zafar Ali Khan, Uday B. Desai Department of Electrical Engineering Indian Institute of Technology Hyderabad, India Email: {yuva,ee12m1029,zafar,ubdesai}@iith.ac.in Abstract—In this demo, we demonstrate integration of cogni- tive radio capabilities, on the fly, with GSM base station (BTS) using OpenBTS architecture. The demonstration has following novel features: • Cognitive sensing of the channel, in use by secondary user, while transmitting, • On the fly shifting to free GSM subchannel without call drop, • Works with existing GSM phones without any modifica- tion in hardware and software. Trials at IIT Hyderabad over the last one year has validated the viability of the system. Keywords—Cognitive Radio, Spectrum Sensing, OpenBTS, GSM I. I NTRODUCTION Over past few decades increase in the wireless technologies and applications like GSM, Bluetooth, Wi-Fi, broadcast TV, etc., has found increase in the usage of available spectrum which has created spectrum scarcity. Future wireless commu- nication focuses on delivering maximum throughput, better coverage, high data rate, high spectral efficiency by dynamic spectrum access. These deliverables could be achieved by effectively using cognitive radio technology which enables sensing of free licensed spectrum belonging to primary user (PU) and utilization of the spectrum efficiently by assigning or allowing ‘free spectrum‘ usage by secondary user (SU). In order to implement cognitive radio (CR) communication technologies, it is necessary to evaluate the practicability of algorithms in real time. It also makes sense to use existing architectures to reduce the implementation time. Thanks to the availability of Software Defined Radio (SDR), it is now easy to modify radio capabilities via software and can be made as potential testbed. Open source hardware platforms like Ettus USRP, Range Networks RAD-1, Rice WARP, Fairwaves UmTRX can be used as off-the-shelf SDR. For robust voice communication, GSM technology is still the easiest and most viable solution. No wonder it is the most successful communication standard in the world covering over 90% of globe population. Inspite of this, it has been repeatedly observed that utilization of GSM spectrum is poor. For exam- ple in New York and Chicago, GSM cell phone frequencies are occupied 45% and 55% of the time, respectively [3]. In this demo we concentrate on building of a CR testbed using existing GSM protocol stack and that aims to provide QoS for both PU and SU. Fig. 1: Cognitive GSM BTS II. RELATED WORK After the emergence of SDR, open source based application based on SDR has seen tremendous growth. Especially, open- source software for telephony like OpenBTS [4], OpenBSC [5], Osmocom BB [6], Openmoko [7] has proved that real time voice or data communication can be done at low cost, providing a viable solution for less dense populated areas like rural and underdeveloped countries. In particular, OpenBTS has revolutionized the GSM telephony making GSM BTS setup possible using SDRs like Range Networks RAD-1, USRP, UmTRX, etc. . Efforts to use OpenBTS as cognitive radio testbed has been reported in [1] and [2]. In [1], cognitive capability is incorporated indirectly by monitoring frame error rate (FER). If the FER is above a certain threshold, it is assumed that the PU is active and the call is abruptly dropped. This scheme suffers from high false alarm rate (because of obvious reasons) and abrupt call termination. In case of [2] frequency selection is done based on GSM whitespace database from PU which doesn’t involve spectrum sensing and on the fly detection and shifting of PU frequency. In this demo, our contribution is focused on making OpenBTS a cognitive BTS (CGBTS) making it smart to detect the PU on the fly and assign the available free frequency to SU without degrading QoS of both PU and SU. III. COGNITIVE GSM BTS CGBTS allows secondary user to make GSM based calls using OpenBTS without call drop and degradation of QoS for primary as well as secondary user. As mentioned earlier, we need to detect the presence of PU in a reliable manner while transmitting on the same frequency. We achieve this by placing the sensing radio just outside the cell coverage of our SU pico 2014 IEEE 11th International Conference on Mobile Ad Hoc and Sensor Systems 978-1-4799-6036-1/14 $31.00 © 2014 IEEE DOI 10.1109/MASS.2014.68 529