K-means grouping dynamic frame slotted ALOHA-based radio frequency identification tag anti-collision protocol for dense IoT applications

Internet of things (IoT) as a paradigm evolution of the internet promises to enable just anything to be globally identifiable, accessible and data exchangeable. This is only possible with IoT enabling technologies of which radio frequency identification (RFID) is key. However, RFID suffers tag collision problem (TSP) because tags use same communication channel to respond to reader queries. EPC Class 1 Generation 2 (EPC C1G2) protocol has been standardized to address TSP for few tag density. Dynamic frame slotted ALOHA (DFSA) is the de facto algorithm used in EPC C1G2. DFSA fails woefully given high density tag deployment like in IoT as issues like poor tag number estimation, a bogus frame size adjustment scheme and tag starvation problem (TSP) persist. The main aim of this research study is to develop an enhanced RFID tag anti-collision media access control (MAC) protocol that can support dense IoT application of RFID. To achieve this aim, this study first developed a tag grouping and counting algorithm using improved k-means clustering. This algorithm assigns a groupID to each tag using tags’ RN16 as basis for clustering and returns an accurate tag estimate to the reader. Secondly, the study integrates a frame size adjustment look-up-table and virtually divides the communication channel (frame) into subframes. Hence, tags are forced to only reply to reader queries using their allocated subframe. Objective three proposes a two-stage RFID tag identification model – tag initialization and tag identification stages and referred same as k-group dynamic frame slotted ALOHA (kg-DFSA) protocol. In tag initialization stage, when the reader is up, it first runs the new tag grouping and counting algorithm and gets accurate tag estimate while each tag is assigned groupID. In the identification stage, the rigorous and repetitive reader handshaking of each tag before tags are read in traditional DFSA is eliminated while a new tag estimation scheme that uses one subframe for tag estimation is developed under an optimal grouping condition. Traditional DFSA was implemented as an M-file in MATLAB. Each algorithm of the objectives were implemented as MATLAB functions that return values to DFSA. 16-bit binary numbers were randomly generated in MATLAB to simulate tags’ RN16. The simulation study involves one reader to numerous tags and is in conformity with EPC C1G2 standards. Monte Carlo simulation was performed under two simulation scenario - scenario I involves 5 to 95 tags while scenario II (IoT scenario) involves 100 to 1000 tags. Results show the proposed kg-DFSA edges DFSA by about 68% in terms of system efficiency, almost 20% in terms of computational complexity, 27.5% in terms of identification time, zero tag estimation error rate, over 6% in terms of tag coverage rate, almost 35% in terms of slot efficiency and 56% in terms of collision rate in IoT scenario. This implies that the proposed kg-DFSA protocol is relevant in enhancing EPC C1G2 protocol to support dense IoT application of RFID.

Text FK 2023 43 - Declaration Form.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (368kB) Text FK 2023 43 - Full Text.pdf Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (16MB) Text FK 2023 43.pdf Restricted to Repository staff only Available under License Creative Commons Attribution Non-commercial No Derivatives. Download (16MB) Official URL or Download Paper: http://ethesis.upm.edu.my/id/eprint/18963

Actions (login required)

View Item