Link Utilization Error Free Operation Stop and Wait

  • Slides: 18
Download presentation

Link Utilization �Error Free Operation �Stop and Wait: �Sliding Window (W=window Size=2 n-1): 2

Link Utilization �Error Free Operation �Stop and Wait: �Sliding Window (W=window Size=2 n-1): 2

Link Utilization �Probability of frame error = p �Stop and Wait Utilization �Probability of

Link Utilization �Probability of frame error = p �Stop and Wait Utilization �Probability of a frame requiring exactly k transmission = pk-1(1 -p) �Expected Number of Transmission for a frame (Nr): �Utilization should be divided by Nr : �Selective Reject Utilization �Exact same idea as stop and wait 3

Link Utilization �Go back N Utilization �Each frame error requires re-transmission of L packets

Link Utilization �Go back N Utilization �Each frame error requires re-transmission of L packets where L>=1 �Total number of frames that should be transmitted if the original frame must be transmitted k times=f(k)=1+(k-1)L 4

5

5

Multiple Access Control Sublayer shared wire or medium 6

Multiple Access Control Sublayer shared wire or medium 6

Multiple Access Links Two types of “links”: • point-to-point – PPP for dial-up access

Multiple Access Links Two types of “links”: • point-to-point – PPP for dial-up access – point-to-point link between Ethernet switch and host • broadcast (shared wire or medium) – old-fashioned Ethernet – 802. 11 wireless LAN shared wire (e. g. , cabled Ethernet) 7 shared RF (e. g. , 802. 11 Wi. Fi) shared RF (satellite) humans at a party (shared air, acoustical)

Point-to-point networks are those in which, when a message is sent from one computer

Point-to-point networks are those in which, when a message is sent from one computer to another, it usually has to be sent via other computers in the network. 8

Broadcast networks have a single communication channel that is shared by all the machines

Broadcast networks have a single communication channel that is shared by all the machines on the network. Shared channel 9

Channel Allocation �Channel Allocation problem: �Single channel has to be shared among multiple users

Channel Allocation �Channel Allocation problem: �Single channel has to be shared among multiple users �Solutions: �Static allocation: user gets a channel for the whole “conversation” �Dynamic allocation: channel capacity allocated for a user is not fixed; depends on usage 10

Static Channel Allocation �Using FDM or TDM to divide channel capacity among N users

Static Channel Allocation �Using FDM or TDM to divide channel capacity among N users �Inefficient bandwidth use: some users may be idle �Users may be denied access even when other users do not use there allocated capacity �Long time service delay T 11

Static Channel Allocation � Delay in a single M/M/1 Queue �Channel can transfer C

Static Channel Allocation � Delay in a single M/M/1 Queue �Channel can transfer C bits/sec �Frames arrive randomly with an average rate of λ frames/sec and the inter-arrival times are exponentially distributed �Frame length values has an exponential distribution with an average length of 1/µ bits. �Mean delay of frames in the queue is given by: �Example: C=100 Mbps, 1/µ=10000 bits, λ=5000 frames/sec, T=? ? sec � Delay if we divide the channel into N sections �Assume we divide the same stream into N smaller streams and serve each of them in a separate queue �There will be N channels with a capacity of C/N bits/sec for each �There will be N streams with an average rate of λ /N frames/sec each �Mean delay of each queue: 12 �Example: If we divide the above queue into 10 sections, ����=2���

Dynamic Channel Allocation �In data communication, the peak to mean ratio of traffic can

Dynamic Channel Allocation �In data communication, the peak to mean ratio of traffic can be very high. �In many applications, users require the channel in random times �Static allocation will be a waste of channel resources �It is best to assign the channel based on user demand �Different protocols for coordination: 13 �ALOHA : pure and slotted �CSMA : carrier sense multiple access protocols �Collision free protocols �Limited contention protocols �Wireless LAN protocols

Dynamic Channel Allocation �Assumptions for analysis allocation algorithms: of dynamic channel �Station model: �N

Dynamic Channel Allocation �Assumptions for analysis allocation algorithms: of dynamic channel �Station model: �N independent stations generating frames within interval Δ t with probability = λ Δ t where λ is frame rate (frames/second) �Single communication channel �Collision occurs simultaneously �Time is when 2 frames are transmitted �continuous, or �slotted (discrete intervals) 14 �Stations may have the hardware to notice: �Collision of frames �The presence of another signal on the channel (carrier sense)

Pure Aloha • • Station transmits as soon as it has a frame If

Pure Aloha • • Station transmits as soon as it has a frame If a collision is detected, it waits for a random time interval and then retransmits 15

ALOHA Efficiency �Notes and assumptions: �Large number of stations �Frames of equal length �Transmit

ALOHA Efficiency �Notes and assumptions: �Large number of stations �Frames of equal length �Transmit time for frame on the channel is normalized to unity �Stations can sense channel collision and re-try transmission if there is a collision 16

ALOHA Efficiency �Analysis: �All stations together, generate an average of S new frames per

ALOHA Efficiency �Analysis: �All stations together, generate an average of S new frames per frame time �If S>1, every frame suffers a collision, so for reasonable operation, 0<S<1 �A station does not send a new frame until the old frame has been transmitted successfully. So, S can be seen as a measure of throughput as well. �Probability of k attempts per frame time by all stations, 17 old and new combined is Poisson with mean G per frame time. �Low load => rare collision => S~G �Increase the load => More collisions => G>S �Throughput is then the offered load, G, times the probability of no collision, P 0 => S=G*P 0

ALOHA Efficiency �Probability of k frames generated per frame-time (t) is (Poisson distribution) �Probability

ALOHA Efficiency �Probability of k frames generated per frame-time (t) is (Poisson distribution) �Probability that there is no frame generated in a given interval t is �No collision if no other frame generated in two frame-times. �Mean number of frames generated during 2 t sec is 2 G 18 so: