This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.
The IEEE 802.11x set of standards, also known as Wi-Fi (Wireless Fidelity - a trademark of Wi-Fi Alliance), are the most known wireless local area network (WLAN) protocols. The initial version of IEEE 802.11 was released in 1997 and clarified in 1999 . The specification defined three distinct physical layers (PHY), the Frequency Hopping Spread Spectrum (FHSS), the Direct Sequence Spread Spectrum (DSSS) and the Infrared (IR). The first two layers are designed to operate over the Industrial, Scientific and Medical (ISM) band of 2.4 MHz, in 13 overlapping selectable channels (Europe), with a bandwidth of 20 MHz each, and data bitrates of 1 and 2 Mbps. The IR is now obsolete.
IEEE released in 1999 the 802.11b and the 802.11a specifications, which they offered improver physical layers over the 2.4 ISM and the 5 GHz bands, respectively [4, 5]. The IEEE 802.11b offers improved PHY over the 2.4 MHz ISM band and theoretical throughputs 5.5 and 11 Mbps using DSSS. The IEEE 802.11a offers throughputs 6/9/12/18/24/36/48/54 Mbps using a 52 subcarrier Orthogonal Frequency Division Multiplexing (OFDM) over the Unlicensed National Information Infrastructure (U-NII) band. Both continue to use a 20 MHz bandwidth, as the legacy 802.11. In 1999 IEEE released the IEEE 802.11g, a major improvement of 802.11b, which uses the same band of 2.4MHz, the same bandwidth of 20 MHz and combines OFDM and Complementary Code Keying (CCK) for backward compatibility .
Fig. 1: Sub-layers and IEEE 802.11 variants
Essentially, the IEEE 802.11x set belongs to a wider IEEE 802.x family, which includes wireless and wired protocols like the IEEE 802.15 Personal Area Networks (PAN) , the IEEE 802.3 Ethernet etc. . The IEEE 802.11x family specifies only the two lower layers: The Physical and the Data Link layer (see Fig. 1). The last comprises of the Logical Link Control (LLC) sub-layer, which is common in several 802.x protocols, and the Media Access Control, which is common among the several 802.11x protocols.
Fig. 2: An Extended Service Set with Ethernet DS.
The IEEE 802.11 standard defines two distinct network topologies, the Basic Service Set (BSS) and the Independent Basic Service Set (IBSS). The first implementation, also known as Infrastructure mode, comprises of several client Stations (STA) associated with a single Access Point (AP), which serves all the uplink (STA to AP) and downlink (AP to STA) traffic . In the second implementation, also known as Ad-hoc mode, every station becomes independent and exchanges traffic with all his neighboring stations. Both WLANs are identified by a unique name, called the Service Set Identifier (SSID), which is emitted as a 6-byte (48-bit) identifier in beacon frames. When two or more APs, which form two or more BSSs, are connected through a Distribution System (DS) like Ethernet, they create an Extended Service Set (ESS), as shown in Fig. 2. The Extended Service Set ID, which is a 32-character identifier (in ASCII), is emitted in beacon frames from all the APs that belong to the same ESS [8, 9].
The IEEE 802.11 MAC sub-layer specification defines two operational modes, the Distributed Coordination Function (DFC), including the RTS/CTS sub-mode, and the Point Coordination Function (PDF). In the DCF, all client stations contend equally to gain access to the wireless channel, using the Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA), a distributed random access protocol. In the PCF mode, the coordination function is located on the AP, and administrates the access to the wireless channel via a polling mechanism. The associated STAs who has data to exchange, reply to AP polls and the coordinator grand them access time according to their demand, without contention. The PCF mode is often called Contention Free Period (CFP) in contrast with the DFC mode, which is called Contention Period (CP).
The following chapter discusses about DCF mode, because it is the legacy mode where the Enhanced Distribution Channel Access (EDCA) mode of the IEEE 802.11e specification is based, to provide QoS extensions to Wi-Fi networks .
1.2 The Distributed Coordination Function
The DCF mode is based on the CSMA/CA protocol. The function of CSMA/CA is similar to the CSMA/CD, which is the base of the IEEE 802.3 Ethernet protocol, with a major difference: The wireless transceiver is incapable to operate in full duplex mode, because it cannot scan the channel while transmitting, so it functions only in half-duplex mode. If another transceiver transmits the same time in the range of the receiver, the collision that occurs will not be detected by both transceivers. The sender assumes that a collision has happened when it will not receive the acknowledging (ACK) message from the receiver. Therefore, the protocol tries to avoid the collisions as much as possible.
The CSMA/CA operates on the MAC layer. When the LLC layer delivers a packet - a MAC Service Data Unit (MSDU) - to the MAC layer, the station scans the wireless channel (Carrier Sensing, CS) for an active broadcast. If the channel is found IDLE for a minimum time period, called the DCF Inter Frame Space (DIFS), the station begins the transmission and the rest of the stations in range become idle until the sending completes. The idle period is equal to Net Allocation Vector (NAV) and is calculated from the Duration field of the transmitting frame. When the destination station successfully receives the frame it sends an acknowledging (ACK) signal back to the sender after a fixed period, called the Short Frame Inter Frame Space (SIFS), and the transaction is completed . Fig. 3 displays the above procedure.
Fig. 3: The basic DCF access method.
If a sending station finds the medium BUSY through a CCA, it defers transmission and enters into Backoff period, which will be explained later in detail. Consequent transmissions are separated with one of the three inter-frame spaces (IFS), according to the priority of the frame exchange. The period between the transmission of DATA and the ACK from the receiver has the highest priority and it lasts SIFS, which is the shortest period. DIFS is the longest period and has the lowest priority. Therefore, stations waiting for DIFS before transmitting are unable to interrupt a successful frame exchange between two stations.
The third IFS is called the PCF Inter-Frame Space (PIFS), it stands between SIFS and DIFS, and is used exclusively by the PCF mode. As mentioned before, the point coordinator function resides on the AP and it polls the individual stations sequentially, with higher priority than the distributed function. The values of the three IFS depend on the specific physical layer that the 802.11x uses, and are multiples of the Slot Time .
Fig. 4: Inter-Frame Spaces relationships.
The slot time is an elementary time period and depends of the physical layer's attributes, e.g. the propagation delay, the transmitter delay, etc. Fig. 4 depicts the IFS relationships. They are:
The 802.11 transceiver performs two types of carrier sensing. On the first type, the physical layer before transmitting it performs a Clear Channel Assessment (CCA) on the wireless channel, and reports the status found to the MAC layer. On the second, when the MAC layer receives a frame, which was not destined for the specific station, it reads the duration field and defers any transaction for the specific period (NAV). The second case belongs to the extend concept of Virtual CCA, which will be discussed more on the 802.11e protocol.
1.3 The Collision Avoidance Function
When two stations try to emit after a successful CCA, a collision will happen on the wireless channel and both signals will be lost. The CSMA/CA tries to avoid such situations, therefore every station waits an additional period after DIFS and either the channel is busy or not. Essentially, the station defers its transmission by a period called Random Backoff, which is an integer multiple of the Time Slot. The value of this period is chosen from a uniform distribution of the interval [0, CW]. The quantity CW is the Contention Window and is a constant number, equal to an initial default number of time slots. For the DSSS mode it is 31. The CW is randomly selected to avoid the situation where two or more stations start transmitting exactly when their timers expire, which it will guide to a new collision.
All the deferred stations, after the DIFS period and the successful selection of a random backoff interval, start to decrement their timers, every time slot, by one until they zeroed. It is obvious that the station with the smallest selected value will expire earlier and therefore will start transmitting first. Then, the rest of the stations pause their timers and enter to DEFER state, which is precisely calculated from NAV. After the completion of the transmission, the stations that hold data enter to DIFS period again and after the expiration they continue to decrement their backoff timers until they zeroed or interrupted one more time by another earlier timeout.
The initial CW is set to a minimum CW value, which is called CWmin. If two stations contend for the access to the channel and they randomly have chosen the same backoff value, their timers will expire concurrently and their transmissions will collide. Then both stations increase their CW to a new value, given by the equation:
The above is an exponential function and it is often referred as Exponential Backoff. If the two counters expire together again, then their CWs will increase one more time and if this continues they will finally reach the maximum CW size, which is the CWmax. Both minimum and maximum values depend on the physical layer protocol. Fig. 5 depicts the exponential increments for the DSSS physical layer, with the seven consecutive increments of 31/63/127/255/511/1023. In the last two steps CWmax remains the same.
The effectiveness of the exponential backoff stands where a doubled CW decreases in half the probability for the two counters to have chosen the same backoff value and to have a new collision. The DCF protocol defines the Retry Limit, which is the maximum number of unsuccessful retransmissions. If that value is exceeded then the frame is dropped. After a successful transmission the CW is restored back to CWmin. The same delay of DIFS plus a random backoff is used again after a successful ACK, if there is another frame to be sent by the same transmitter. This is called Post Backoff and performs a form of fairness among the transmitting stations, because it allows the rest of the paused timers to continue to countdown until zeroed, and the respective station to have access on the channel earlier [8, 9].
Fig. 5: The exponential backoff
When there are many stations in range that exchange data, the number of collisions increases and the frame drop rate increases, as well. In fact, the protocol will increase the length of the contention window, perhaps up to the maximum limit. This reduces the number of collisions and consequently the frame drop rate. Unfortunately, this has a drawback because every frame is finally transmitted after a number of unsuccessful retries, introducing a significant amount of latency on the transmission, with the long term of reduced throughput and low channel efficiency.
1.4 A sample of the DFC operation
The following figure 6 is an illustrated sample of a typical DCF operation. It supposes that there are three stations in range, operating in ad-hoc mode. Station 3 has data to send and perform CCA, during the DIFS period. Channel is found IDLE and there are not any active backoff counters, so station 3 starts transmitting immediately. During transmission, the channel is found BUSY by the other two stations. Actually, stations 1 and 2 receive data from the upper layers so after the DIFS period they choose a random backoff value, 10 and 19 respectively from the window [0, 31]. Station 3 has more data to send and after the first ACK it enters to DIFS and chooses a new random backoff, equals to 25. It is clear that station's 1 timer will expire first, so station 1 will seize the channel and start transmitting. Station 2 and 3 pause their times at 9 and 15, respectively and enter to DEFER state during the activity of station 1.
Fig. 6: A sample of the DCF function.
After the completion of station's 1 transmission, all stations delay DIFS, station 1 chooses a new backoff value, which is 9, and station 2 and 3 continue to countdown their timers until station's 2 expire concurrently with station's 1. There it happen an unavoidable state, a collision between station's 1 transmission and station's 2. Both signals get lost and stations 1 and 2 double their CW values and choose a new value from [0, 63] windows, which are 52 and 43, respectively. After the DIFS period all counters start to countdown and station's 3 twice paused counter reaches zero after 6 time slots. Then station 3 seizes the channel and transmits the rest of its data.
1.5 The RTS/CTS Function
Wireless protocols that use the CSMA/CA, encounter a functional problem called the Hidden Node Problem. It happens when station B receives data from station A, which is in range, and a third station C, which is out of range of station A, starts to send data to station B. The result is that data from station A collide with data from station C on the receiving station B and the collision avoidance fails. The problem is reduced by introducing the RTS/CTS handshaking function, where the sending station after the DIFS period, sends a Request to Send (RTS) signal to the receiving station (see Fig.7). All stations in range, which are not addressing any data, interpret the frame by extracting the duration field, and defer their activity up to the end of the transaction (including the SIFS period and the ACK).
Fig. 7: The RTS/CTS function.
Station B after receiving the RTS signal (plus the SIFS period) it replies station A with a Clear to Send (CTS) signal. Similarly, all near stations that receive the CTS, hold their data according to duration field and eventually they do not interrupt the transmission until the entire handshaking (including SIFS intervals and final ACK) will complete. During the RTS/CTS handshake, the SIFS periods prevent any transmission from other stations because the DIFS period (where station performs CCA), is longer that SIFS and the BUSY state of the channel will be detected .