Transmitter Location Estimation for Radio Environment Map Construction using Software Defined Radio Mustafa Tugrul Ozsahin Computer Networks Research Laboratory (NETLAB) Bogazici University, Turkey Email: tugrul.ozsahin@boun.edu.tr Tuna Tugcu Computer Networks Research Laboratory (NETLAB) Bogazici University, Turkey Email: tugcu@boun.edu.tr Abstract—This work focuses on the practical implementation of transmitter location estimation algorithms using software defined radio devices for the purpose of Radio Environment Map (REM) construction, which is used to determine primary user location and available channels for Cognitive Radio. To this end, a testbed is also set up and technical and physical details of this testbed are explained. The location of an immobile transmitter is estimated using the collected signal strength data from multiple receivers by applying one known technique: LocatIon Estimation based (LIvE) REM Construction Method, and one proposed technique: exhaustive search on the discrete map. The results of these estimations are explained and compared. I. I NTRODUCTION Cognitive Radio (CR) concept is proposed to better utilize the frequency spectrum by providing opportunistic spectrum access. To achieve this goal, it uses spectrum sensing function to determine the unused frequency bands, which is mainly done by the measurement capable devices in the network. Sensing might be non-cooperative where CR users detect the primary transmitter signal independently through their local observations, or it might be cooperative where information from multiple CR users are utilized for primary user detection. Sensing and detection is not always sufficient for estab- lishing a complete environmental awareness for a CR network. The Radio Environment Map (REM) concept is proposed as an enabler for CR networks to provide comprehensive environ- mental awareness [1]. Essential functionality of a REM is the construction of dynamic interference map for each frequency at each location of interest. REM collects spectrum measurements from the nodes with spectrum sensing capabilities. These nodes might be the CRs in the network or dedicated spectrum sensors [2]. It generates a spectrum usage map by processing these data. In this work, we study practical implementation of trans- mitter location estimation for REM construction purpose. We set up a testbed, and taking sufficient amount of measurements, first we estimate the channel path loss properties for the open field. Then, we investigate the possibilities of estimating the unknown location of the transmitter using the sensing data coming from multiple receivers. We test and validate the usability of a known transmitter location estimation technique of a REM construction method given in [3], and compare its results with exhaustive search based methods by practical implementation. II. SPECTRUM OCCUPANCY & REM The most important function of the CR is spectrum sensing because the origin of the problem is the sparse use of the spectrum and the purpose is to exploit the unused portions (spectrum holes) and make better use of them. Sensing can be classified as cooperative and non-cooperative, according to the information usage and synthesis scheme used to decide on a PU transmitter’s presence. A. Cooperative Sensing If one SU decides based on its own local measurements, it is considered as non-cooperative sensing. If the sensing information from multiple SU receivers are considered to make a decision, it is considered as cooperative sensing. Detection with cooperation among multiple SUs is more accurate compared to non-cooperative sensing. However, coop- eration also increases the energy consumption of the network. Since the devices in a CRN are generally mobile devices with local power sources, optimizing the tradeoff between the accuracy and the energy consumption is also important for feasibility of the system [4]. Cooperation of SUs for successful sensing limits the num- ber of channels that can be sensed. As a result, the utilization is affected by the number of sensing nodes. Thus, a subset of the channels can be selected considering the traffic requirements, and the SUs can be assigned to those channels for sensing. In [5], the problem of channel and user selection is considered for cooperative sensing. B. Radio Environment Map Instantaneous information of detected PU signals gathered from the individual CR devices is not always sufficient for establishing a systematic way of learning available spectrum bands. The occupancy of some licensed bands may vary from time to time, and this usage scheme can be periodic, or aperiodic. For such reasons, establishing a more systematic knowledge base for spectrum occupancy information is essen- tial [6].