Radio Layer Physical Layer Computer Science Essay

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

In this paper we explore the background and architecture if Bluetooth. The overview and functionality of Bluetooth layers. The architecture of Bluetooth Baseband layer. Different routing schemes and comparison between their efficiency and power consumption.


Bluetooth is a wireless technology designed for short-range communications systems intended to replace the cables. Its key features are robustness, low power and low cost. Bluetooth which was presented in Feb. 1998 by five promoters, Ericson, Nokia, IBM, Toshiba and Intel; after introducing Bluetooth formed a group called, the BLUETOOTH SIG. The purpose of Bluetooth was to create a cheap wireless connection between different portable devices. The name Bluetooth was taken from Harald Blatand, a Danish Viking king. Bluetooth is low power consumption technology that is why it is integrated in many mobile and other network devices. The communication range of Bluetooth connection is 10-20m [1].


Bluetooth model is identical to generic OSI model. It is composed of layers: Radio Layer, Baseband Layer, Logical Link Control Adaptation Protocol and Application Layer [2].

Figure 1: [1]


The first layer is Radio Layer. The Radio Layer describes the physical radio system, identical to the physical layer of OSI model. The main functions of this layer are interference performance, out-of-band blocking, intermodulation characteristics etc [2] [3].


The next layer is Baseband Layer which is responsible for transmission and reception of data packets, error detection and encryption. The Bluetooth baseband layer is the physical connection or transmission path between two or more points that use the fundamental frequency of a communication system. The primary and secondary communicate with each other using time slots. The time slot between two intervals is same that is the settling time (625 microseconds). The time slot is the interval in which a sender send frame to secondary or secondary send frame to primary. Communication is between primary and secondary, secondary to secondary communication is not possible. The baseband layer is responsible for the conversion of data packets into assigned type of RF packets. Different types of RF packets hold information in different types of data fields. Asynchronous packets contain address and control information. Synchronous packets do not contain address information. The baseband layer uses the proper modulation type that is GMSK, QPSK and 8DPSK to convert the digital bits into an RF signal. [4]. The baseband layer senses the received RF power level and adjusts the transmitted RF power level. On this channel the information is transferred through the packets. Packets can cover one time slot time or up to five slots time. Bluetooth protocol uses a combination of packet and circuit switching. This layer can support asynchronous data channel and synchronous voice data channels. Due to this layer Bluetooth can support point-to-point connection and point-to-multipoint connection [2] [3].

2.2.1. TDMA:

Bluetooth uses a form of TDMA that is called TDD-TDMA .TDD-TDMA is a half duplex technique in which sender sends data and receiver do not receive data on same time; however, different hops uses different directions. If TDDA has only one secondary then it is called Single-Secondary Communication, and is easy. Even slots are for Primary that is 2, 4, 6... And odd slots are for secondary that is 1, 3, 5…. .Multiple-Secondary Communication is a little more complex as more than one secondary is involved. The primary uses even slots but the secondary uses the next odd numbered slot for addressing, if packet was sent in previous slot [3].


The physical channel is defined automatically by seventy nine or twenty three randomly generated RF channels. Each piconet hop addressing is unique and it is determined by Bluetooth device address of the master; phase in the hopping sequence is determined by the master's clock.

The channel is divided into time slots according to the hop frequency. All units participating in piconet are synchronized (time and hop) to channel. These slots are numbered according to the Bluetooth clock of master. In time slots the masters and the slaves' exchanges packets. The master transmits packets in even slots. And the slaves receive packets in even time slots. The packets should be aligned by time slots. Packets transmitted may be extended up to five time slots [3].


Different types of links can be established between masters and slaves. Two common types are Synchronous connection-oriented (SCO) link and the other is Asynchronous connectionless (ACL) link. The SCO link is point-to-point link between a master and a slave. The master in SCO reserves the slots periodically. And ACL is a point-to-multipoint link between a master and slaves in a piconet. In ACL master don't need to reserve particular slot for any slave [2] [3]. SCO LINK:

The SCO link is point to point connection between a master and a slave on reserved intervals. SCO link is used for fast transmission rather than error free transmission. There is fixed direction for hop frequency, as it is point to point connection. So if a packet is not reached at slave or is damaged, it will be lost. There is no feedback system. SCO links is mostly used for audio transmission or video transmission where delay is not required [3]. ACL LINK:

The ACL links are used where delays are not important but the integrity of data is very important. In this case slots are not reserved and master can interact with any slave. It provides packet switching between master and slave. Between master and slave only one ACL link can be established. If packet is corrupted the packet is retransmitted to ensure data integrity. If there is no data to send then no polling is required and no transmission takes place [3].


The baseband layer interprets the bit arrived from higher layers as the bit has to be transmitted in air. Generally, packet header and payload header fields are at baseband layer. The data in piconet is transmitted in the form of packets. Each packet is composed of an access code, a header and the payload. The numbers of bits reserved to each part are also indicated. The access code and header are of fixed size that is seventy two bits and fifty two bits, respectively. Whereas, size of payload bits is variable; ranging from zero to 2745 bits [3].

Figure 2: [3] ACCESS CODE:

Each packet starts with access code of 72 bits if packet header is followed by it. Else it has length of 68 bits. The purpose of access code is synchronization; identification and DC offset compensation. All packets in a piconet are preceded by same access code [3]. PACKET HEADER:

The header contains LC information and has six fields. AM ADDR (3-bit address), TYPE (4-bit type code), FLOW (1-bit flow control), ARON (1-bit acknowledge indication), SEQN (1-bit sequence number), HEC (8- bit header error check).

This makes the total header of 18 bits size [3].


This layer can do multiplexing, it receivers data from senders side, make frames and send data to baseband layer. Another function of this layer is addressing between them, assure quality of service level [2].


Bluetooth has two types of networks: Piconet and Scatternet. Piconets are defined as two or more

units transferring information. Numerous piconets with

overlapping coverage form a scatternet [4].


Figure 3: [6]

The Baseband design has been chosen with standard Generic architecture. The main advantage is that basic management functions are performed by an external microcontroller. The microcontroller clock is set to 10MHz. We do TDM (Time Division Multiplexing) by synchronizing the clock. The clock from sender's side must be synchronized with the clock at receiver's side. We know that baseband layer extracts data from L2CAP layer, which is being sent in the form of packets [4].


Figure 4: [5]The Bluetooth access scheme used is based on polling that is one of the device acts as master and one act as slave. MAC is a Master driven Time Division Duplex (TDD) system. Full duplex transmissions, Time Division Duplex (TDD), is supported in Bluetooth baseband layer which divides each second into 1600 time slots. The transmission of a Baseband packet usually uses a single slot but, may last up to five consecutive timeslots. A master is a device that establishes the connections to distant device, slaves. A slave cannot establish any connections; it will just respond to the connections from the master device. Therefore, traffic in the Piconet is mainly controlled by Master; and slave transmits a Baseband packet if it receives a packet previously from Master. The master discovers slave devices and their services and is capable of connecting to multiple slaves, and can handle active connections simultaneously. A master can connects up to seven slaves at a time. In Bluetooth connection between two devices it does not matter which is slave and which one is master. However, if we wish to connect more than two devices in the same session it is possible that we will have to consider how the Bluetooth master / slave roles impact upon the connection setup. Because once a master connected seven slaves, this master cannot connect any more slaves. The solution to this problem is that a slave in one piconet can act as master in other piconet [5].

However, it is not an easy to establish such connections. When the master has no data to send, it sends no payload data called a POLL packet. Then, the slave responds to POLL packet by sending NULL packet back. There are four operational modes for Bluetooth Slave: Active, Sniff, Hold and Park. In Active, Sniff and Hold modes, a Slave has a 3-bit address (AM ADDR) which it uses to communicate with the Master. In the Park mode, a Slave instead of AM ADDR has 8-bit addresses (PM ADDR, AR ADDR). Thus, Park mode can connect more than seven Slaves. However, in the Park mode a Slave cannot transmit or receive data. It has to listen to periodic Master transmissions for synchronization. A Slave is inactive for a fixed time interval in Hold Mode, while in Sniff mode a Slave wakes up after regular intervals to listen for Master transmission. A slave unit working in Hold or Park mode in a piconet can join another piconet but in different mode. In sniff mode the slave is active for some interval, so it can visit any other piconet in the inactive interval called sniff slots. Bluetooth is a fast frequency hopping system and Master in the piconet determines the hopping sequence. This assures that each piconet has a unique frequency hopping sequence [6].


As name refers all slaves are organized in single piconet. Park mode is used to accommodate more than seven slaves. Periodically slaves are parked and unparked depending if in active mode and timestamp. Each Slave remains unparked for the sometime interval. This model is and does not require inter-piconet communication. However for large number of Slaves, the time a Slave remains parked can be considerable [7].


Slaves in several overlapping piconets form scatternet model. In this model different Communication groups are formed from different piconets. This allows instantaneous communication in different Communication Groups and hence leads to higher throughput, low delays contrary to single piconet model [6] [7].

In this model, identifying a Communication Group is insignificant task. Communication Group can be identifies by a traffic control pattern. Firstly, a single piconet is formed for all slaves. And destination addresses are observed in slave packets and the Master determines the flow of traffic and hence different communication Groups can be identified. A single piconet is split into several piconets. Traffic distribution changes with time and by observing the addressing outside the piconet, master decides when the slave should move to another piconet. Well this is not an efficient method for determining Communication Groups. The piconets formed on the basis of Communication Groups will have less inter-piconet communication. The multi inter-piconet communication results in interference which is a major drawback [9].


An efficient scheduling policy depends upon the state of the queues at the Master and Slaves, the traffic entrance process at queues, and the packet length distribution at Master and Slave. The factors of interest are the system throughput, packet delays, fairness and the packet drop probability. The two objectives of throughput and fairness can conflict since the fairness might have to be sacrificed for throughput and vice-versa. As mentioned above there is alternative time slots for Master and Slaves. This implies that if a forward slot is assigned to a Master then the following reverse slot is assigned to the respective Slave. Thus, scheduling occurs in the Master- Slave pair. Further, the task of scheduling depends on the Master; this makes Bluetooth a Master driven standard. There could be wastage of slots in the TDD scheme, if only one of the Master or the Slave has data to send, a slot gets wasted. Further, the fairness issue is complicated if Master and Slave do not, at the same time, have data to send. Due to above reasons, Round-Ribbon is extensively used due to low throughput with TDD based MAC protocols, and make sure no fairness. New and efficient scheduling policies are required to solve above mentioned issues. These policies should be simple to implement and cost effective [9].

We purpose "Master-Slave Queue-State-Dependent Packet Scheduling Policies". Masters and slaves are assigned different states according to master-slave pairs in the queues. If master-slave pair has to send data then it is denoted by 1 and if date is to be received then it is denoted by 0. This leads s to four different states that are 1-1, 1-0, 0-1 and 0-0. Clearly if master has to send data and if slave has to receive data then it has 1-1 state. If master sends data but slave don't receive data then the state is 1-0. If master has no data to send but a slave receives data then the state is 0-1. And at the end if both master and slave don't send and receive data respectively then state is 0-0. One bit is required to send information. Master has the information of queue at the slave's side. In the priority policy (PP), a higher priority is given to the Master-Slave connections in the 1-1 state over pairs in 0-1 or 1-0 states. The 1-0 and 0-1 Master-Slave pairs have equal priority.2 Master-Slave pairs in 0-0 state are not scheduled. Note that the PP policy achieves a higher throughput than pure round robin policy since the connections in 1-1 state are given a higher number of slots. As a result, there is less wastage of slots since fewer slots are given to 1-0 and 0-1 pairs. The important parameter in this policy is the priority P given to the 1-1 connections over 1-0 or 0-1 (1-0 or 0-1 connections get a priority of 1). We perform round robin scheduling among all the Master- Slave connection pairs that are in 1-1, 0-1 or 1-0 states [9].


In this paper we are focusing on the efficiency and techniques of different algorithms. The piconet and scatternet polling is based on different algorithms.


In round robbin system, the master polls the slaves consecutively. Each slave is just allowed to send one packet per cycle. Master can assign packets of different bandwidths and types to slaves in Round Robbin Scheduler. So users can have higher bandwidths on demand and this is the main feature of this polling [2] [9].


In this scheme the master continues polling until the slave until the master queue and the output slave queue is emptied. This assures not only the packets at the beginning of the cycle but also those cycles generated during the cycle are served before the master moves to next slave. This algorithm favors slaves by generating packets at maximum rate [2].


This is the combination of two above mentioned schemes. The main idea of this scheme is to poll slaves which often have nothing to send. The slaves are in two states active and inactive. The polling starts with master who moves all slaves into active state and begins sub cycling schemes. In a polling sub cycle Round Robbin, all active slaves are polled in round ribbon fashion. The master has little knowledge of arrival process at slave's side. The job of master is scheduling the uplink and the downlink. So a feedback system is needed for uplink and downlink connections during polling the slaves. Depending upon the feedback the slaves are moved between different states. A slave goes into inactive state only if sender has no data to receive and master has no data to send. The slave is in active state if master has information to send [2].


This algorithm is based on two main ideas; it is necessary to avoid Null transmission from master to slave and important is to maintain the fairness of Round Robbin scheme. This can be possible if master to slave polling is under the consideration of master's knowledge of traffic to and from the slaves. In this algorithm we assume that the master assigns a separate queue for each slave participating in communication. The master has knowledge of slaves' state (i.e. active or inactive). We assume that master gets the knowledge from the feedback during polling the slave. Thus master can only predict if slave will send NULL from slave's behavior in the previous polling cycles. The goal is to do efficient scheduling algorithm that is not dependant on implementations; the slaves should not be bounded to adopt the same header fields as assigned by the master. Here we define slave to master directions as uplink and from master to slave as down link. The polling of slave and master causes coupling in transmission, therefore it is not possible to remove slave from the cycle without blocking the transmission from slave. We can decouple scheduling in transmission in uplink and downlink by double polling [10].


Baseband Layer is responsible for transmission and reception of data packets, error detection and encryption and is the physical connection or transmission path between two or more points that use the fundamental frequency of a communication system. We see that the polling takes place from master to slave. The master can connect up to seven slaves. There are several routing algorithms the pioneer algorithm and most efficient is Robbin Ribbon algorithm, which controls fairness. The other algorithms were developed to control more data rate, multi-directional polling and error free transmission.