Research And Design Of Anti-Collision Algorithm Based On ALOHA For RFID System

RFID (Radio Frequency Identification) is a widely used wirelessautomatic identification technology. It implements the bi-directional andwireless communication by means of radio signals in the space. RFID systemnot only has a large capacity of access data, but also possesses thecharacteristics of multi-target recognition, high identification speed, low costand power consumption, which is the core technology in the newresearch——the IoT (Internet of Things).Multi-target recognition, the main advantage of RFID, meets the needfor entire acquisition of mass information in the IoT. Multi-target recognitionis a significant reason of RFID’s development. It is a key prerequisite toenhance this advantage that researching reliable and efficient anti-collisionmechanisms. Nowadays, most of the anti-collision algorithm researches focuson the deterministic algorithms based on binary search and the probabilisticalgorithms based on ALOHA. Binary search algorithms are bound by tag ID’s length. ALOHA algorithms, by contrast, are more suitable for IoT’sapplications in which tags with large quantity and dense distribution areprocessed.EPC C1G2protocol is universal, high-speed in identification, standardscompliant. It conforms to the EPC standards. EPC C1G2protocol adopts aframed slotted ALOHA named Q-algorithm. Q-algorithm’s frame length is2Q,which can be adjusted through Q. Each tag randomly generates an integerbetween0and2Q-1,and load it in the slot counter. Tags with0in their slotcounters respond immediately to the command. In this dissertation,Framed-Slotted ALOHA Algorithm with Continuous Slot Prediction andDynamic M-ary Hashing was proposed against defects in original protocol.This algorithm is composed of multiple mechanisms including DynamicM-ary Selection, Continuous Slot Prediction and Frame Length Adjustmentbased on Continuous Slot Prediction.EPC C1G2protocol only provides the processing scheme of the currentslot. With the Continuous Slot Prediction mechanism, this proposed algorithmcan manage the tags in continuous slots and skip several slots at a time. Itaccurately estimates collision tags’ quantity with the state information of continuous slots. Therefore, this algorithm can reduce massive invalid slots.The proposed algorithm adopts the frame length adjustment schemebased on Continuous Slot Prediction. This algorithm can adjust Q based onthe condition of continuous idle slots and continuous collision slots, makingup the inflexibility of Q adjustment in original Q-algorithm. By the analysisof continuous slot states mathematical model based on Markov chain, theproposed algorithm provids the threshold of Q adjustment in the ContinuousSlot Prediction mechanism. In this situation, Q adjustment is faster and moreaccurate, making frame size as close with the number of tags as possible.The algorithm efficiency further improves through the Dynamic M-aryHashing mechanism and the design of Additional Frame Ladd. The proposedalgorithm adds the additional frame (Ladd) to solve collision tags separately.This design can not only avoid repeated collisions among collision tags andothers, but also estimate accurate quantity in additional frame. The collisiontags are evenly distributed in additional frame through Dynamic M-aryHashing mechanism. There are only a few collision slots and idle slots inadditional frame. If collision happens, the algorithm solves the collisionsinstantly with binary selection. The proposed algorithm improvements focus on using Continuous SlotPrediction mechanism in each frame to accelerate skipping invalid slots andadgust Q, solving collisions and hashing tags with Dynamic M-ary Hashingmechanism. This algorithm ensures that the system works in the best state.Experiments with MATLAB were conducted including frame lengthadjustment speed, transmission overhead and identification delay, throughput,ect. Simulation results prove that the proposed algorithm can increase thesystem’s throughput, reduce delay and overhead. Compared with otheralgorithms, superiorities of the proposed algorithm are more outstanding indense circumstance with large quantity of tags. The proposed algorithm isappropriate for RFID terminals in the IoT.