Wireless LAN (IEEE class of protocol suite)

Transcription

1 Wireless LAN (IEEE class of protocol suite) In these notes, I will explain the MAC protocols used in constructing wireless LANs using the IEEE class of protocol suites. Specific protocols, such as the , will be discussed because these have been standardized and these are used world-wide. At this point, it is important to note the following assumption made in the design of the CSMA/CD protocol to construct a wired LAN: signal from any computer in a network can reach all computers in the network. In a wired LAN, this assumption must hold for computers to be able to reach each other and for the collision detection mechanism to work. However, in a wireless LAN, the above assumption may not hold. That is, signal from a computer may not reach all the computers in the network. Hence, at the MAC level, we do not go for collision detection. Another consequence of signal not reaching all the computers in a network is that computers in a network may not be able to communicate directly. For two computers to be able to communicate, service of another computer in the network or a fixed infrastructure, namely, an Access Point (AP), is used. C: Computer, AP: Access Point C Radio range of the AP Access Point IEEE protocol C Router To the rest of the network AP Figure WL1: Elements of a wireless LAN (WLAN) A wireless LAN using an Access Point (AP) is shown in Figure W1. An AP is the main part of the infrastructure to support a wireless LAN. An AP is connected to a router, which in turn is connected to the rest of the network and the Internet. One of more APs 1

2 may be connected to the same router as shown in the figure, or different APs may be connected to different routers. The APs as well as user computers are equipped with an identical radio interface commonly known as the IEEE standard interface. The IEEE standard specifies both the PHY (physical) and the MAC layers for constructing a wireless LAN. The PHY layer defines the air interface between two IEEE equipped devices and the MAC layer defines the medium access control protocols to access the shared medium. (There are a variety of the IEEE protocol, such as a, b, g, and e. Protocols such as the , a, b, and g differ only in their PHY part, thereby supporting different levels of bandwidth, whereas all their MAC protocols follow the same principle. On the other hand, the e MAC, which has been designed to support multimedia applications, is different from the above protocols. In these notes, we will not discuss the e protocol for lack of time. However, once one understands the MAC protocol, it is easier to understand the more complex e protocol.) Different modes of operation of IEEE MAC protocol Before discussing the details of the IEEE MAC protocol, it is useful to understand the different modes of operation of the protocol as classified below. Modes of IEEE MAC Distributed Coordination Function (DCF) mode Point Coordination Function (PCF) mode With Hand-shake Without Hand-shake Figure W1: Different modes of operation of MAC protocol in a WLAN PCF mode of operation In this mode of operation, an AP acts as the central controller for all the computers lying within the radio range of the AP. The AP decides who transmits when. Thus, there is no 2

3 contention among computers for medium access. The AP can, but not necessarily, follow a round-robin policy to assign time slots to individual computers, thereby guaranteeing a certain amount of bandwidth to each computer. Because of this guarantee of allocation of bandwidth, this mode of operation can support real-time traffic. However, if the amount of traffics generated by different computers widely vary, this mode is not suitable. This is due to the fact that the AP may allocate bandwidth to a computer which may not have much data to transmit. Finally, the IEEE standard says that implementation of the PCF mode of operation is optional. DCF mode of operation In this mode of operation, the presence of an AP is not essential. For example, computers can communicate among themselves using the DCF mode of operation without the assistance of an AP. In such a case, we say that the computers have formed an ad hoc network, where there is an absence of infrastructure, such as APs. However, to construct a wireless LAN with the capability to connect to the Internet, it is essential to have an AP which is connected to a router as shown in Figure W1. Whether or not there is an AP, in the DCF mode of operation all the computers compete for medium access, because there is no central controller if there is an AP, it does not act as a central controller. Because of contention among computers for medium access, there is no guarantee of bandwidth and there is no bound on message delay. Thus, the DCF mode of operation is not suitable for real-time traffic. Rather, data delivery is made on a best effort basis the network makes an effort to deliver data across the wireless medium, but there is no guarantee of bandwidth or bound on delay. Finally, the IEEE standard says that implementation of the DCF mode of operation is mandatory, that is, the DCF mode of operation must be supported by a network card claiming to implement IEEE Note: We have said that the PCF mode is optional, whereas the DCF mode is mandatory. If an AP implements both the modes, the following is a scenario in which both the modes can be used alternatively. For example, the network can run in the PCF mode for T1 seconds, followed by the DCF mode for T2 seconds, followed by the PCF mode for T1 seconds, and so on. In this manner, all, or, some, computers get a guaranteed bandwidth during the PCF mode, whereas all the computers compete for bandwidth in the DCF mode. Those computers without data to be transmitted will not compete for medium access during the DCF mode, thereby allowing those computers with data to enter the contention phase. PCF DCF PCF DCF PCF.. Time Figure W2: Alternative use of PCF and DCF modes 3

4 DCF with hand-shake In this mode, a sender obtains permission from an intended receiver before transmitting a packet to the receiver. This is achieved by the sender transmitting a Request To Send (RTS) packet to the intended receiver, and the receiver granting permission by transmitting a Clear To Send (CTS) packet back to the sender. The intended receiver may refuse to grant permission by remaining silent, i.e. not transmitting a CTS packet back to the sender. The sequence of RTS and CTS control packets is known as the process of hand-shake. An intended receiver can refuse to receive a packet from the sender for a variety of reasons, such as lack of buffer and lack of interest in communicating with the sender, to name a few. This mode of operation is preferred where the network traffic is high or if the packet to be transmitted is a long one. In either case, if hand-shake is not used, the packet will collide with another packet at the receiver with high probability. Thus, it is recommended to use the hand-shake mode of DCF when traffic is very high in the network or a packet is very long. The flip side of using hand-shake is the extra overhead of exchanging RTS/CTS control frames. Essentially, before transmitting a long packet, a sender not only informs the intended receiver about the upcoming transmission, but also both the sender and the intended receiver inform their neighbors via the RTS and CTS control frames about the upcoming transmission so that the neighbors do not initiate their own transmissions while the packet is being transmitted and acknowledged. (The idea of data packet acknowledgement at the MAC level will be explained in a later part of this section.) DCF without hand-shake In this mode of operation, a sender does not take permission from an intended receiver before transmitting a data packet. That is, RTS/CTS control frames are not exchanged between a sender and a receiver before the sender transmits a data packet. Rather, the sender just transmits the packet when certain medium sensing conditions are satisfied. These conditions will be explained in a later part of these notes. Thus, there is no coordination between a sender and a receiver prior to a data packet being transmitted. This lack of coordination increases the probability of packet collision. If data traffic is low or packets are short, then this mode of data transmission avoids the overhead due to hand-shaking. Note: The two modes of DCF operation, namely with hand-shake and without handshake, are not mutually exclusive. Rather, both coexist, and the two modes can be selected on a packet basis. For example, the dotrtsthreshold variable in the MAC management database can be used to choose a certain mode of operation. The dotrtsthreshold variable is initialized with an integer value representing a threshold packet length in bytes. If the length of a data packet is equal to or greater than the dotrtsthreshold value, then the MAC layer chooses the with hand-shake mode. On the other hand, if the packet length is smaller than the dotrtsthreshold 4

5 value, then the without hand-shake mode is chosen to transmit the data packet. This situation has been further explained in Figure W3. Figure W3 shows that a wireless LAN can alternatively operate in the PCF and DCF modes, and within the same DCF mode duration, a computer can successively transmit two data packets in the without handshake and the with hand-shake modes. PCF DCF PCF DCF PCF.. Time With hand-shake Without handshake }mode of operation of the same computer Figure W3: Alternative use of PCF and DCF modes and consecutive appearances of with hand-shake and without hand-shake modes in the same DCF period. Problems in a wireless LAN There are three problems in a wireless LAN, which are not found in wired LANs. These problems are as follows: - Hidden terminal problem - Exposed terminal problem - Inability to detect collision. Hidden terminal problem Assume that all the computers have identical radio ranges. This assumption is not necessary for the problem to occur. Rather, the problem becomes more frequent in the absence of this assumption. Thus, for simplicity of our explanation, we make this assumption. Let there be three computers A, B, and C as shown in Figure W4. Their relative locations and radio ranges have also been shown in Figure W4. The figure shows that A and B are within each other s radio ranges, and B and C too are within each other s radio ranges. Thus, the direct communications that can happen are between A and B and between B and C. However, A and C are outside each other s radio ranges. Now, assume that C is transmitting to B as shown in the figure. Since C is far away from A, A is unaware of the fact that there is an on-going transmission between B and C. Because of this unawareness, A can potentially start its transmission to B or another computer (not shown in the figure), thereby disturbing C s transmission to B. This disturbance happens because C is hidden from A. Thus, a hidden terminal problem occurs when a computer (A) starts its transmission while being unaware of a far-away located computer s (C s) 5

6 transmission to A s neighbor (B). In other words, the problem can occur if two computers (A and C) have a common neighbor (B), but the two computers can not hear each other s signals. Because users, and, hence their computers, are mobile in a wireless LAN, the hidden terminal problem occurs in a wireless LAN. Suitable measures are taken in the IEEE MAC protocol to address this problem. Tx A B C Figure W4: Illustration of the hidden terminal problem Exposed terminal problem Referring to Figure W5, assume that the pairs of computers (D, A), (A, B), and (B, C) are Neighbors in the sense that D and A are within each other s radio ranges. Let A be transmitting a data packet to D. Because of the broadcast nature of this transmission, the same signal also reaches computer B. However, B is not an intended receiver of A s signal. We also know that C is not receiving A s signal. Thus, it is possible for B to start transmitting a data packet to C without disturbing the ongoing transmission from A to D. However, B does not initiate its transmission, because it is unaware of D s location. B does not know if its transmission to C will cause a collision at D. Hence, B does not initiate its transmission while being exposed to A s signal. This is called the exposed terminal problem. In other words, a computer does not start transmitting if it senses that there is an ongoing transmission. Researchers are investigating this issue so that network performance is improved by allowing an exposed terminal to transmit subject to the condition that it does not disturb an ongoing transmission. D Tx A B C 6

7 Figure W5: Illustration of the exposed terminal problem Inability to detect collision Ideally, a sender should detect collision at a receiver, because it is the collision at the receiver that matters. In a wired LAN, collision is detected at the sender s end by assuming that signals from all computers can reach all other computers. However, this assumption does not hold in a wireless LAN as it has been explained in the context of the hidden terminal problem. Another reason for the difficulty in not being able to detect collision is that many wireless devices use half duplex transceivers to simplify transceiver design. That is, a half-duplex wireless device turns on its transmitter and receiver in an alternating manner. As an example, GSM phones use half-duplex transceivers, and this is facilitated by skewing the up-link (transmitting) channel and the down-link (receiving) channel by three time slots so that the transceiver can switch its mode from a transmitter to a receiver and vice versa. Thus, in a wireless LAN, because of the difficulty in detecting collision, MAC protocols do not rely upon collision detection. Rather, the principle of collision avoidance is utilized. Transmit condition and virtual carrier sensing Before a computer can transmit anything data or control packets it is essential to detect idleness of the medium. In a wired LAN, it was sufficient to detect idleness by sensing the physical medium. However, in a wireless LAN, because of the hidden terminal problem, it is not enough to sense the medium to know whether or not the medium is idle. In addition to carrier sensing at the PHY level, the WLAN MAC protocol uses the concept of virtual carrier sensing to tackle the hidden terminal problem. In virtual carrier sensing, a computer monitors the transmission of all control packets that it can receive, utilizes the duration information contained in those packets, and infers whether or not the medium is idle at its intended receiver. Thus, the medium is said to be idle if the PHY level carrier sensing mechanism detects no carrier and it is inferred from the virtual carrier sensing mechanism that the intended receiver within its radio range is not receiving data. 7

8 Transmit condition Medium is idle Absence of carrier detected by the carrier sensing mechanism at the PHY level. and The virtual carrier sensing mechanism tells that the medium is free. Figure W6: Condition for medium idleness Virtual carrier sensing is implemented by using two concepts as listed below: - Control frames (or, packets) o Request-To-Send (RTS) frame o Clear-To-Send (CTS) frame - Network Allocation Vector (NAV): This is a scalar, integer variable held by each computer. Suppose that a computer A wants to transmit data to another computer B. Before transmitting the data frame, A transmits an RTS frame to B. The major fields of an RTS frame are: A s and B s addresses and a duration field. The duration field tells B the length of time for which the medium is likely to remain busy due to A s data transmission and B s acknowledgement. An RTS frame is a point-to-point frame in the sense that only B should respond (a CTS frame) to the RTS frame. Also, an RTS frame is a broadcast frame in the sense that all those receiving this frame are supposed to extract the duration field out of the packet to update its NAV, but the unintended receivers do not send back a response (a CTS frame). By transmitting an RTS frame, computer A achieves two objectives as outlined below: - Computer A requests permission from B. - Computer A informs all its neighbors of a likely data transmission within a certain time interval so that the neighbors do not transmit anything within the said interval. If, or when, B receives the RTS frame, it decides whether or not to receive a data frame from A. In case B wants to receive a data frame from A, it responds back to A by transmitting a CTS frame. Contained in the CTS frame will be A s address and the 8

9 duration for which the medium is likely to remain busy around B. All computers, except A, receiving the CTS frame extract the duration value and update their NAV variables. Neighbors of A will receive A s RTS and neighbors of B will receive B s CTS. All computers, except the pair A and B, receiving an RTS or a CTS shall not initiate a transmission in the period denotes the duration fields of the RTS/CTS frames. Computers update and interpret their NAV variables as follows. - Each computer has its own NAV with an initial value of 0. - NAV represents for how long the medium is likely to remain busy. - The unit of time information in NAV is microseconds. - NAV is decremented by 1 with the elapse of each microsecond of real time. - When NAV = 0, it is no more decremented. - NAV is updated using the duration field in a received control frame as follows: o NAV = Max(NAV, duration value received in a control frame) - NAV = 0 means no pre-announced transmission is going on in the vicinity. - Idle medium is defined as absence of carrier AND NAV = 0. Timing relationship between RTS and CTS The hand-shake mechanism using RTS and CTS control frames has been illustrated in Figure W7. In Figure W7 we assume that computer A wants to send a data frame to computer B, C is a computer within the radio range of A, and D is a computer within the radio range of B. We also assume that D is out of range of A, and C is out of range of B. The four frames, namely RTS, CTS, DATA, and ACK, are separated by a time interval known as Short InterFrame Space (SIFS). The idea behind this spacing is to maintain a logical association between these frames. Because of propagation delay and the delay involved in computing a response to a frame, there is a need to have an allowed time gap between logically related frames. For example, when a station, say, B, has successfully received an RTS, it should be given an opportunity to respond back with a CTS without any disturbance from other computers. This is achieved by allowing B to make a decision within SIFS interval and send back a CTS. It may be noted that no other computer within A s and B s transmission range should transmit a packet during the SIFS gaps in Figure W7. This is achieved by having the computers sense the medium to be idle for more than the SIFS interval. (These intervals are known as PIFS Point InterFrame Space and DIFS Distributed IFS.) An important thing in Figure W7 is the concept of acknowledgement (ACK) sent from B to A. If B receives the DATA frame successfully, it sends back an ACK to A. It is useful to recall that there is no such ACK mechanism in the CSMA/CD protocol for Ethernet. The reason for including an ACK at the MAC level in wireless LANs and not having it in the MAC protocol for wired LAN is that the bit error rate (BER) in case of high quality wired LAN is very low, whereas the BER in a wireless environment is comparatively very high. For example, the BER on a fiber link is as low as whereas it is 10-5 on a 9

10 wireless link. If the concept of ACK is not used at the MAC level in a wireless LAN, upper level protocols will have to handle the resulting packet loss. Because of the longer timeouts involved in upper-level protocols and the possibility of loss of retransmitted packets from the upper-level protocols, handling packet loss at the upper-level is much more expensive than the cost involved at the MAC level where the probability of packet error is high. However, if the probability of packet error is extremely low at the PHY level, handling packet error is left to the upper-level protocols due to the extra cost of running the ACK mechanism at the MAC level. Value of duration in RTS A RTS SIFS DATA Value of duration in CTS B SIFS CTS SIFS ACK C Value of NAV of C D Value of NAV of D Time Figure W7: The hand-shake mechanism using RTS/CTS Note: Two computers, by exchanging RTS and CTS frames, inform their neighbors to not initiate a transmission during the period reserved by the RTS/CTS frames. Thus, the neighbors refrain from initiating their own transmissions during the said period and help in avoiding potential collisions. Hence, the RTS/CTS mechanism works to achieve the goal of collision avoidance. 10

11 DCF mode of operation with hand-shake In the following, we give a flow-chart representation of the DCF protocol with handshake. We have already explained the basic concepts used in the protocol, such as the idea of RTS/CTS. We will present the protocol in two parts in the form of a sender s behavior (Figure W8) and a receiver s behavior (Figure W10). 11

12 F: a new data frame to be transmitted i = 0, CW = CW min NAV = 0? End of backoff Idle medium for DIFS interval? No Random Backoff CTS is received ACK is received Yes Send an RTS Start a timer Cancel timer Send DATA (F) Start a timer Timeout Timeout Wait for fairness to others Cancel timer Wait for a random interval i = i+1 CW = CW min *2 i (At some point, CW saturates at CW max.) i: Retry count, CW: Contention Window CWmin: Minimum value of CW (typical value is 32) CWmax: Maximum value of CW (typical value is 256) DIFS: Distributed Interframe Space SIFS < DIFS Important note Figure W8: Hand-shake mode of DCF operation (Sender s behavior) The random backoff protocol (the dotted box in Figure W8) The backoff mechanism in the WLAN protocol works differently from the backoff mechanism in CSMA/CD protocol for Ethernet as follows. - Choose a random number in the range [0, CW]. A variable, called Backoff Time Counter (BTC) is initialized with the chosen random number. The unit of time represented by the BTC variable is denoted by aslottime, where 12

13 aslottime accounts for propagation time and transceiver switching time (switching time between transmitter and receiver and vice versa.) - BTC is decremented by 1 if the channel is idle for aslottime. - Stop decrementing the BTC if the medium is busy. - Resume decrementing the BTC after finding the channel to be idle for DIFS interval. Thus, whenever the channel is found to be busy, decrementing of BTC is done after finding the channel to be idle for DIFS interval immediately after the busy period ends. Once the channel is idle for at least DIFS interval, subsequent decrement is done for each idle period of aslottime interval. (See Figure W9 for an illustration.) - BTC = 0 means that the backoff process has finished. A computer remains in the backoff state for at least BTC time units of real-time and during this period the medium is idle. In Figure W9, we give an example of how BTC is decremented by computer B. We assume that the channel remains busy due to activities of computers A and C. Here, the focus is on how BTC decrements and not on who should transmit when. The random initial value of BTC is assumed to be 5. Note: The reader may note an important relationship between SIFS and DIFS in the form of SIFS < DIFS. In the DCF mode, a computer senses the medium to be idle for at least a DIFS period. Thus, even if the computer finds the medium to be idle for SIFS period between frames in the sequence {RTS, CTS, DATA, ACK}, it does not initiate a transmission in order to not disturb an ongoing communication. Thus, the relationship SIFS < DIFS is very important. A recommended equality relationship between SIFS and DIFS is: DIFS = SIFS + 2*aSlotTime (This relationship will be made clearer in a later part of these notes.) 13

14 Ch. Busy (A) Ch. Busy (C) Time B DIFS DIFS X X X X BTC = If the channel is busy, it has to remain idle for DIFS interval for BTC to be decremented by 1. X = aslottime If the channel is idle for at least DIFS interval, it has to remain idle for aslottime interval for BTC to be decremented by 1. Figure W9: Illustration of how BTC decrements from a random initial value of 5. 14

15 Receive an RTS Remain silent. NAV = 0? No Receive a DATA frame Yes Channel is idle for SIFS and the computer is ready to receive? No Ch. idle for SIFS? Yes No Yes Send an ACK Send a CTS Note: The above two fragments of flow-charts can be easily merged. Figure W10: Hand-shake mode of DCF operation (Receiver s behavior) 15

16 DCF mode of operation without hand-shake The DCF mode of operation without hand-shake is a special case of the DCF mode of operation with hand-shake. In this mode, RTS and CTS frames are not exchanged between two computers before the sender transmits its DATA frame. A receiver must respond to a correctly received DATA frame with an ACK frame. The idea of NAV is still used in this mode of operation. This is because of the fact that even if a computer does not use the hand-shake mode, it must update its NAV if it receives an RTS or a CTS frame from another computer. A network card may implement both the modes of DCF operation and dynamically decide what mode to use on a frame basis. Note: A computer may wish to broadcast a DATA frame to all its neighbors. By broadcast, we mean all the neighbors are the DATA frame s intended receivers. For a DATA frame to be broadcast, we need to remember two points as follows: - A DATA frame is broadcast in the DCF without hand-shake mode. - Receivers of a broadcast frame do not send back ACK frames. Thus, though broadcast operation is a desirable one, there is no guarantee of its reliability at the MAC level, i.e., there is no guarantee that all the neighbors of the sender will receive a copy of a broadcast frame. A broadcast frame is distinguished from a point-topoint DATA frame by using an appropriate addressing mechanism in the header of DATA frame. PCF mode of operation In the PCF mode of operation, an Access Point (AP) is a key element in a WLAN. The AP controls the channel by alternating between the PCF mode and the DCF mode. For an AP to be in full control of the channel, it grabs the channel as follows: - The AP senses the medium to be idle at the beginning of a contention-free (CF) period for a PIFS (PCF InterFrame Space) interval. The PIFS interval has an important relationship with the other two intervals, namely, SIFS and DIFS, in the form of: SIFS < PIFS < DIFS. This is how we interpret the above relationship. Frames in the sequence {RTS, CTS, DATA, ACK} maintain a gap of SIFS. By having SIFS < PIFS, the AP lets an ongoing transmission to be completed before grabbing the channel. By having PIFS < DIFS, we make sure that another computer is prevented from acquiring the channel, thereby giving priority to the AP to grab the channel. This happens because in the DCF mode of operation, a computer must observe the channel to 16

17 be idle for at least DIFS interval. In terms of equality relationships, recommended values of SIFS, PIFS, and DIFS are as follows: PIFS = SIFS + aslottime DIFS = SIFS + 2*aSlotTime - If an AP finds that the medium is idle for a PIFS interval, it transmits a Beacon frame. A Beacon frame contains a CFPMaxDuration field (CFP: Contention-Free Period.) This value communicates the length of the CF period to all stations. Essentially, this value tells the user computers the maximum duration for which the AP will be controlling the medium. Computers receiving a Beacon frame update their NAV value with the CFPMaxDuration value. Consequently, user computers are prevented from taking control of the medium until the end of the CF period. This is because for a user computer to access the medium, first its NAV must reach 0. - After transmitting a Beacon frame, the AP waits for at least one SIFS interval and next transmits one of the following frames: o DATA frame o CF Poll frame o Data+CF Poll frame o ACK frame o CF End frame The above frame types have been explained in the following. - DATA frame: o This frame contains user data from the AP to a particular station. o Upon correctly receiving the DATA frame, a user computer sends back an ACK frame to the AP after an SIFS interval. o If the AP does not receive an ACK, it can retransmit the frame after a PIFS interval. o An AP can also transmit broadcast and multicast DATA frames to all the users or a subset of the users, respectively. o Broadcast and multicast frames are not ACKed. - CF Poll frame (explained in Figure W11): o An AP sends this frame to a computer granting the computer permission to transmit a single DATA frame to any destination the AP or another user computer. o Receiver of the above DATA frame sends back an ACK to the polled computer (This is the computer that sent out the DATA frame.) o If the polled computer has no DATA frame, it must send a null DATA frame. o If a polled station does not receive an ACK, it can not retransmit its data frame unless it is polled again. 17

18 AP User 1 User 2 AP User 1 CF Poll CF Poll SIFS DATA DATA SIFS ACK SIFS SIFS ACK The polled user sends data to another user. The polled user sends data to the AP. Figure W11: CF Poll frame - DATA + CF Poll (See Figure W12.) AP User 1 User 2 DATA+ CF Poll ACK SIFS DATA ACK SIFS The polled user receives data from the AP and sends data to another user. Figure W12: DATA+CF Poll frame 18

19 - CF End: This frame identifies the end of the CF period. This is sent by an AP when one of the following happens: o The AP has no data to transmit and no computer to poll. This can happen before the pre-announced CFPMaxDuration value expires. Thus, the AP can prematurely terminate its CF period by transmitting a CF End frame. When the user computers receive this frame, those reset their NAV to 0. o The CFPDurRemaining time expires. This variable is initialized with the value of CFPMaxDuration that was broadcast in a Beacon frame. As time passes, CFDurRemaining is decremented and its value represents the remaining time for which the AP will be in control of the channel. A computer joining a WLAN with an AP In this part of the notes, we explain how a computer joins a wireless network with an AP. (There is no need for such a protocol in a wired LAN, because by plugging in a new computer to a wired LAN, your computer automatically becomes a part of the LAN.) A station can join a WLAN in one of two ways: - by passive scanning - by active scanning Passive Scanning: A station listens to each channel for a specific period of time. (A channel can be considered as a certain carrier frequency. All the computers and the AP transmit on the same carrier frequency. If the radio ranges of two APs overlap, they use different channels.) Initially, a computer does not know if there is a nearby AP and the channel used by the AP. Thus, it has to listen to each channel for a certain interval. Listening to a channel means waiting to receive a Beacon frame transmitted on the channel. After the computer receives a Beacon frame, the computer can negotiate a connection with the AP by proceeding with the authentication and association processes. Active Scanning: In this mode of joining a WLAN, a computer sends a Probe frame to an AP by using the broadcast mechanism, and waits for a Probe Response from the nearby AP. A Probe Response frame identifies the presence of a network. Once a computer receives a Probe Response frame, it can negotiate a connection with the AP by proceeding with the authentication and association processes. 19

20 Comparison of CSMA/CD MAC for LAN and CSMA/CA MAC for WLAN Finally, we compare the CSMA/CD and CSMA/CA protocols for Ethernet LAN and wireless LAN, respectively. The comparison will focus on the important concepts involved in the design of those protocols. CSMA/CD CSMA/CA (Ethernet (Wireless Features LAN) LAN) Comments Collision Yes No In wireless networks, in general, Detection it is not possible to detect collision. ACK of No Yes In a wireless network, it is important data frames to ACK data frames because of the high BER (bit error rate). Possible use of a central controller No Yes An AP in a WLAN is a central controller. Multi-mode No Yes PCF and DCF (with/without hand-shake) Data transmission Sensitivity to No Yes Short packet: DCF without hand-shake data frame length Long packet: DCF with hand-shake Fairness Yes Yes Achieved using a random wait after successfully transmitting a data frame. Rate of High Low The low rate in WLAN is due to the transmission unreliable characteristic of wireless medium Real-time No Yes An AP in the PCF mode can support traffic support (best effort) time-sensitive traffic. 20