IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 1, NO. 11, NOVEMBER 2002 1 Expowave: An RFID anti-collision algorithm for dense and lively environments Nikolaos Konstantinou, Member, IEEE Abstract—This paper analyzes and proposes Expowave, a distributed algorithm for the scheduling of an RFID reader network. The behavior of the algorithm is presented in detail, and its performance is evaluated through a set of simulation experiments. It is demonstrated that the algorithm constitutes an efficient approach to the reader anti-collision problem, especially in dense and lively environments. Index Terms—RFID, Reader, Anti-collision. I. I NTRODUCTION D URING the latest years, we are witnessing ubiquitous computing becoming more and more a part of our every day lives. RFID tags and sensors are constantly gain- ing popularity, to the extent that the materialization of the Internet of Things vision is constantly evolving towards its wider adoption. Research and Development in RFID has made applications possible in various domains such as person identification [1], retail stores [2], [3], asset tracking [4], etc. RFID tags are becoming cheaper and cheaper and, compared with other identification technologies such as Datamatrix and Barcodes, they seem to be gaining ground since their price, that has always been the major problem since the inception of the idea, is constantly decreasing while the benefits that occur from their use are increasing. Since its inception, the basic idea of functioning of an RFID tag is backscatter [5]: An RFID reader, also referred to as an interrogator, broadcasts a signal in his interrogation zone. If an RFID tag finds itself in this interrogation zone, it backscatters (i.e. transmits back) a signal containing its unique identification number. The signals that are transmitted and received by the readers and the tags are in a specific frequency and, as such, various problems arise when lots of RFID readers are placed close to each other, forming a densely covered area. Interference between two readers can occur even when these readers’ interrogation zones do not overlap [6]–[8]. In this paper, we study this problem, we propose a novel approach for RFID reader anti-collision and we compare it experimentally, through simulation experiments, to the ap- proaches described in the bibliography. The proposed algo- rithm is proved to be of higher capacity in dense RFID reader networks. Practical use cases of such algorithms include cases when an area needs to be covered with RFID readers in a way that objects moving through are correctly and efficiently identified. Manuscript received June 13th, 2010; revised September 13th, 2010. This work was partially supported by the FP7 Project ASPIRE. The author is with the Athens Information Technology (AIT). The problem in these cases can be simulated as a graph, where each node represents an RFID reader and each edge represents the ability of a collision to occur between the two readers. Thus, the reader anti-collision problem is expressed as: ”how can we organize the system in order to minimize node collisions and achieve the maximum throughput”? The paper is structured as follows: Section II presents the approaches that have been proposed in the literature and Section III describes in detail the proposed algorithm. Section IV presents the experiments that were conducted in order to measure the algorithm’s efficiency and Section V concludes the paper with our final observations and remarks. II. RELATED WORK IN READER ANTI -COLLISION MANAGEMENT First, we need to note in dense and lively RFID environ- ments, two types of reader-to-reader interferences can occur. The first is the frequency interference, also called reader-reader collision and it occurs when two or more readers communicate on the same frequency at the same time. The second type of interference is the tag interference, also called reader-tag collision and it occurs when two or more readers attempt to communicate with a particular RFID tag at the same time. Reader-tag collision protocols can be binary tree-based, as the work presented in [9] or aloha-based [10]. The results presented in this paper provide a solution to the former case. In particular, we do not attempt to resolve interference problems but rather to avoid them by scheduling the readers to communicate in different time periods. The reader anti-collision problem is widely studied in the bibliography. Many approaches to its solution have been adapted from the sensor network research domain, as in [11]– [14]. Among the most notable, and one of the first reader-reader collision avoidance algorithms is Colorwave [15]. Colorwave considers an “interference graph” over the readers, where an edge between two readers means the probability of a collision when transmitting simultaneously and it tries to randomly color this graph in order for each pair of readers to have different colors, each color representing a time slot. The solution presented in this paper builds on top of Colorwave, by inheriting its key characteristics and improving its properties. In [16], the authors suggest coloring the interference graph using k colors, where k is the number of available channels. If the graph is not k-colorable using their suggested heuristic, then the authors suggest removal of certain edges and nodes from the interference graph using other heuristics which consider the size of the common interference regions between neighboring readers.