Research On Throughput Improvement Algorithm Based On Channel Quality Estimation And Retransmission For Bluetooth System

As a typical short-range wireless communication specification, Bluetooth has beenwidely used in a big variety of fields. It has many advantages such as low power, smallsize and convenient to communicat with each other. A Bluethtooth system may performnot well enough in data throughput so even as to cause reduction of the throughput whencommunication quality is not very good. The leading reason of this matter is that theadvantages of Bluetooth mentioned above on the other hand constrain its throughput, ortechnology defined in the Bluetooth specification is not state-of-the-art. Obviously, it isimportant to explore some methodologies to improve throughput between two Bluetoothdevices in bad channel quality condition.This paper firstly introduced the Bluetooth protocol specification, then discussed thefunction of each layer in the procotol stack in detail, especially the data transmissionmechanism in baseband layer and L2CAP layer. Other improvement methods in bothbaseband layer and L2CAP layer were also summarized in this paper. Simulationexperiment was done to illustrate the advantage and the drawbacks of each method aswell. According to the existing problems, the author gave out improvement strategy inboth baseband layer and L2CAP layer.On baseband layer, adaptive packet selection strategy was a sort of approach whichdid not change the Bluetooth procotol stack. But this approach assumed that the channelquality was already known to the system which was in fact unknown in thecommunication of the two Bluetooth devices. Although some literature gave out methodsto change paceket type according to the retransmission times, but most of the researchhad low efficiency.To solve this problem, an adaptive packet selection strategy based onSignal-to-Noise Radio (SNR) estimation was presented in this paper. The receiving sideof Bluetooth system used two-order and four-order moments estimation algorithm toestimate the channel quality, thus the receiver could know which packet type was the bestfor the channel and told this packet type to the transmitting side throuth PDUs in L2CAP layer. This strategy could supervise the channel quality in real-time. Simulation provedthat this strategy could control the estimation errors within0.2dB by using only a fewBluetooth baseband packets.On L2CAP layer, the go-back-N ARQ mechanism could lead to throughputdropping attribute to a lot number of retransmission when the channel quality was not sogood. Selective repeat ARQ mechanism could greatly increase the throughput, but thisstrategy needed enormous buffer size in the receiver and added a sequence number todata frame to discriminate the frame because it might be segmented. So the Bluetoothdevice had to spend extra expense to achieve this strategy.To solve this problem, a dynamic transmission window control method waspresented in this paper. The expression of Bluetooth’s data throuthput under go-back-Nmechanism was put forward, which proved that the change of the transmission windowdid have the function of improving the data throuthput. Furthermore, this expressioncould be used to determine the size of the transmission window. Simulation resultsproved that the throughput of advanced go-back-N ARQ mechanism had improvd datathroughput greater than the previous Bluetooth retransmission mechanism. Moreover,this strategy brought slightly expense to Bluetooth system compared with selective repeatARQ mechanism, so it had more application worth.In summary, methods to improve the data throuthput of Bluetooth system arepresented in both baseband layer and L2CAP layer in this paper. The strategy inbaseband layer is proved to be more efficiency and more accurate than existing methods.The strategy in L2CAP layer can improve the data throughput with little expense addedto the system. Both strategies make no change to the Bluetooth procotol stack, so theyare not restricted to be used to applications.