| Bluetooth devices work in the unlicensed 2.4GHz ISM (industry science and medical) band, whose signal is influenced by multi-path and interfered by other various systems, such as wireless local area network (WLAN), micro-wave oven, wireless phone, and so on. Adjacent Bluetooth device can also lead to throughput of Bluetooth piconet falling. So it is necessary to investigate the throughput of Bluetooth piconet in different transmission conditions and analyze why noise, multi-path, and interference make the throughput of Bluetooth piconet fall and know how much the throughput falls, so as to direct practice and bring forward the methods of coexistence.The typical transmission channels of Bluetooth devices are additive Gauss white noise (AWGN) channel, multi-path fading channel, the channel in which interference of other Bluetooth devices exists, the channel in which interference of WLAN exists. This paper investigates three typical channels, not including the last one, because of not enough time.The investigation methods of throughput in AWGN and multi-path fading channel are similar. Therefore they are reported in one chapter. The condition of Bluetooth channel is often changed. Average receiving SNR can reflect the channel condition in some way[4]. Valenti and so on investigated the packet retransmission probability and the throughput math model of Bluetooth link in AWGN and Rayleigh channel first. Golmie and so on analyzed interference between Bluetooth device and 802.11 device in detail and brought forward solving method[10]. Pasolini and so on investigated the relation between average SNR and packet error rate[14].Kleinschmidt and so on brought forward polling algorithm of Bluetooth piconet in Nakagami-m fading channel, which used channel condition estimation[8]. Sarkar and so on brought forward math method to calculate maximal throughput under the condition of channel known[15]. But the transmitting performances of new ACL packet of Bluetooth 2.0+EDR are not investigated much. So they need to be researched in detail.In a Bluetooth piconet, the Master essentially controls the channel. Due to an absence of coordination between independent Master while accessing the wireless medium, devices will encounter high packet interference if several piconets are simultaneously operating in the same area. Study of packet interference gains importance because if affects throughput of a piconet. Therefore,it has been attracting the attentions of several researchers[28]-[30][32]-[36]. El-Hoiydi[28]has given a probabilistic model of interference of single-slot packets in a homogeneous cluster of piconets, i.e., all piconets are either of 79-hop type or 23-hop type. Naik et al.[34]have given a generalized model of interference of single-slot packets in a heterogenous cluster of piconets. A heterogeneous cluster means all the devices in some piconets are of 79-hop type and all the devices in the rest of piconets are of 23-hop types, not including a piconet where the Master is of the 79-hop (or, 23-hop) type and a Slave is of the 23-hop (or,79-hop) type. All papers assume packet slot rate is 1. But in fact packet slot rate is less than 1. Because not all of the time every packet occupies is used for transmitting data. There is about 259μs used for physical switch. Thus if different packets overlap in the time when there is no data, they will not interfere each other. So previously analyzed results are not fit for practice. In this paper we consider the switch time.First, the paper tells the background, meaning and actuality of this study. And main work is introduced.Second, wireless layer specification, baseband specification and wireless transmitting environment of Bluetooth are analyzed and studied. Third, throughput of Bluetooth piconet in AWGN channel and Rayleigh channel is analyzed respectively. At first bit error rates of three modulation methods of Bluetooth specification 2.0+EDR in AWGN channel are studied. Then packet retransmission probability is analyzed. Throughput and time delay are calculated on this base and emulations are made with Matlab. According to the results of emulation we give a adaptive packet choosing method, which can make throughput maximum at any average SNR. The analyzing method of throughput of Bluetooth piconet in Rayleigh channel is similar to AWGN channel. The difference is that the SNR pdf is needed to calculate packet retransmission probability. In similar way we also give the adaptive packet choosing method in Rayleigh channel. At last packet performances in AWGN channel are contrasted to Rayleigh channel.Forth, packet interference and aggregated throughput in a cluster of Bluetooth piconets under different traffic conditions are studied. First of all in this part, the reason and condition are analyzed. Then packet interference is studied from receiving power, time and frequency. According to analysis, calculation and emulation of receiving power, we know C/I of Bluetooth device in the area is easily less than threshold if there are several Bluetooth devices in the area of 10m. Hence, we consider when frequency and time overlap, interferences between devices come into being in a cluster of Bluetooth piconets. Our model is based on the idea of a probabilistic graph. Our model donsiders the following aspects of data transmission: (1) symmetric and asymmetric traffic ; (2) a cluste of homogeneous piconets using a mixing of 1-slot,3-slot,and 5-slot packets;(3) synchrony and asynchrony between packets transmitted in different piconets. Then several transmission modes are defined, such as symmetric and asymmetric 3-slot and 5-slot packet transmission. Probabilistic graph interference model is defined as this. Let G=(V,E)be a probabilistic graph,where V is a set of nodes and E is a set of edges of G. Let N=|V| denote the number of nodes in G. Each node in G represents a piconet. The weight of an edge between two nodes vi and vj is denoted by p. The quantity p represents the probability of a packet in one piconet, say, vi, being interfered by a packet in another piconet, say, vj. Let X N be a random variable representing the number of nodes in G with no incident edges. Thus, X N is a measure of number of packets not being interfered. The expected value of X N, denote by E ( XN), gives the average number of packets that are not interfered. Interference is calculated by computing the value of p and giving an expression for E ( XN). In the course of analyzing and calculating the probability of packet collision under different traffic conditions, graph and table method is summarized, which is more convenient and more exact. Then an expression of E ( XN) is given. Throughput and packet loss probability are computed on the base of E ( XN). From emulation graph we can know that the packet loss of a synchronous system is always lower than that of an asynchronous one and the packet loss under asymmetric traffic is likely lower than that under symmetric traffic. From graph of throughput we know if the piconets operate synchronously, it produces higher aggregated throughput under either symmetric or asymmetric traffic, compared with the unsynchronized system. It also allows us to accommodate more piconets, say around 60, in an area to reach the maximum aggregated throughput, compared with the unsynchronized one, which can accommodate only around 40 piconets to reach its maximum aggregated throughput. In case the number of Bluetooth devices in the area is more than 320, try to form the cluster of piconets with as smaller number of piconets as possible. In general, it is not a good idea to accommodate more than 50 piconets in a cluster.At last, we make a conclusion for the paper and make a prospect of next work . |
PDF Full Text Request