Single shared broadcast channel, if two or more than two nodes simultaneously transmit, will interfere with each other (interference) conflict (collision): Nodes receive two or more signals at the same time → Receive failed! The MAC protocol uses a distributed algorithm to determine how nodes share channels, that is, when decision nodes can transmit data. It must be based on the channel itself, communication channel sharing coordination information. No out-of-band channels are used for coordination.
- Channel partitioning (channels partitioning) MAC protocol
Tdma:time Division Multiple AccessTDM divides time into timeframes (timeframe) and further divides each timeframe into n slots (slots) per site at each timeframe, occupying a fixed-length timeslot (length = packet transfer time); timeslot idle (idle) 6 site lan,134 Transport grouping, 256 idle
fdma:frequency Division Multiple AccessThe channel spectrum is divided into several frequency bands (frequency bands) each site allocates a fixed frequency band, does not conflict but the channel utilization may not be high, no transmission band idle 6 site LAN, 134 band transmit data, 256 band Idle.
Cdma:code Division Multiple AccessEach user is assigned a unique m-bit sequence (chipping sequence), where "0" is denoted by "1", "1" is denoted by "+1". Each user chip sequence mutually orthogonal (orthogonal) each user uses the same frequency carrier, uses the respective chip sequence to encode the data, encodes the signal = (raw data) x (the chip sequence) such as sends the bit 1 (+1), then sends own m bit chip sequence such as sends the bit 0 (-1), Then send the code chip sequence of M bit chip sequence of the counter-code receiver received is the user's stack vector, with the sender of the chip sequence and received the encoded signal to find the inner product can be decoded
- Randomly access (random access) MAC protocol
When a node is sending a packet, it uses the channel all data rate R to send the packet, conflict occurs when there is no prior node coordination between two or more nodes, and therefore needs to be defined: How to detect the conflict, how to recover from the conflict (such as by delay retransmission)
time Slot Aloha ProtocolAll frame sizes At the same time are divided into equal length slots (each timeslot can transmit 1 frames), the node clock synchronization node can only send frames at the beginning of the time slot, when the node has a new frame in the next time slot (slot) sent if 2 or more than 2 nodes in the same time slot to send a frame, The node detects a conflict if there is no conflict: the node can continue to send a new frame in the next timeslot if the conflict: the node is re-transmitting the frame at the next timeslot with a probability p, until the Success Advantage: (1) A single node activity can transmit data continuously at full channel rate (2) highly decentralized: only synchronous timeslot (3) is simpleDisadvantages:(2) The time slot (2) when the conflict is wasted there is an idle time slot (3) where the node may be able to detect a conflict at much less than the packet transfer time, but there is no point in detecting the conflict because the timeslot Aloha protocol only allows the data frame to be sent at the start of the timeslot (4) requires clock synchronization hypothesis: Each node in each timeslot is sent with probability p for a given node, the probability of a frame being sent in a time slot = P (1-p) N-1 the probability of successfully sending a frame for any node = NP (1-P) N-1 maximum efficiency: To obtain the maximum 1-p for NP (N-1) p* for many nodes, NP * (1-p*) N-1 when N approaches infinity, the maximum efficiency can be =1/e≈0.37
non-time slot Aloha protocolNo clock synchronization required, simpler when a new frame is generated, the probability of sending the conflict immediately increases: when the frame is sent at t0 time, the frame conflict-prone time zone that is sent in [T0-1, t0+1] is [T0-1, t0+1], twice times the timeslot Aloha Protocol p (the given node successfully sends the frame) = P (this node is sent) *p (no other nodes send frames during [T0-1, T0]) *p (no other nodes send frames during [t0, t0+1]) = P (1-p) N-1 (1-p) n-1= P (1-p) 2 (N-1) Select the optimal p, when N approaching infinity limit =1/(2e) ≈0.18, less than the time slot aloha worse
Carrier-listening multi-access Protocol CSMA (Carrier Sense Multiple access) protocolCarrier Listen: Listen channel (carrier) before sending frame: Channel idle: Send full frame channel busy: Delayed send 1-stick to CSMA: with probability p=1 always adhere to the monitoring channel non-stick CSMA: Do not adhere to the monitoring signal, wait for a period of time to listen to the P-stick The CSMA conflict may still occur: the signal propagation delay, or at the same time the end-to-end channel propagation delay of transmitting the data frame broadcast channel, the carrier listening node can not hear another node in the network to start the transmission of the opportunity to begin transmission, B has begun to transmit, but due to signal transmission delay, D can not hear, Therefore also the transmissions began. However, even if a conflict occurs, the data frame must be sent out and the channel resources will be wasted.
Carrier-listening multi-access Protocol CSMA/CD (CSMA with collisiondetection) protocol with Collision detectionApply to Ethernet Collision detection: When a transport node is listening for this channel while it is being transmitted. If it detects that another node is transmitting an interfering frame, it stops transmitting the wired LAN is easy to implement: measuring signal strength, comparing transmit signal and receiving signal wireless LAN is difficult to achieve: the received signal strength is submerged in the local emission signal intensity can detect a conflict in a short time (superimposed signal detected), the post-conflict transmission is aborted, Reduce channel wasteNetwork bandwidth: R bps Data frame min. Length: Lmin (BITS) signal propagation speed: V (m/s) data frame before sending complete, a must receive a conflict from B L/R≥2DMAX/VLMIN/R = 2dmax/v may have some delay time, RTT > d/v , the maximum propagation delay between 2 nodes in the rttmaxtprop= LAN is lmin/r= ttrans= the longest frame transfer delay, for 10Mbps, Transmission of the maximum length of the Ethernet frame time is approximately 1.2msTprop approaching 0 or Ttrans approaching ∞, the efficiency approaching 1 far better than Aloha, and simple, decentralized!
Carrier-listening multi-access Protocol CSMA/CA (Csmawith Collision Avoidance) protocol to avoid collisions802.11 Wireless LAN, do not like CSMA/CD, side send, edge detection conflict! Wireless channel is difficult to achieve all possible conflicts cannot be heard: Hidden station, signal fading802.11 send side:(1) If the monitor hears that the channel is idle for difs time, the entire frame is sent (no conflict is detected at the same time) (2) If the channel is busy, start the random Backoff timer countdown when the channel is idle, when the timer expires, the sending frame (3) If no ACK is received, increase Plus random backoff interval time, repeat (2)802.11 receiving end:If the frame is received correctly, after delaying the sifs time, send an ACK to the sender (due to a hidden station problem) MAC/CA basic idea: Allow the sending side "reserve" channel, instead of randomly send data frame, use small reservation frame to avoid long data frame conflict channel idle difs time , the sending side first uses CSMA to send a very short RTS (Request-to-send) frame to BS RTS frames that may still conflict with each other (but RTS frames are very short) BS broadcast a CTS (Clear-to-send) frame as a response to the RTS, CTS frames can be received by all nodes to eliminate hidden stops affecting the sending side can send data frames, other nodes delay sending
- Take turns accessing the MAC protocol
Polling ProtocolThe primary node polls each node in a circular manner (sending a message to the node, telling it that it is capable of transmitting most of the frames) slave nodes sending data (invited and having data to send data) disadvantage: (1) polling data frame polling overhead (2) The primary node must poll each inactive node in turn, Active node waits for polling presence Delay (2) Single point of failure problem
Token delivery ProtocolA control token (a small special frame) is passed from one node to the next node in turn. When a node receives a token and it has some frames to send, it holds the token and sends the maximum number of frames, otherwise it immediately forwards the token to the next node. Cons: (1) token cost (2) token loss, you must call some recovery steps to return the token back to the Loop (3) The token is passed in turn, and the node must wait for the token to have a wait delay (4) A single point of failure
Computer network-Link layer (2) Multiple Access Control protocol (multiple Access controls protocol)
