Outline

• More Sliding Window Protocol
• In-Class Exercise
• Ethernet

Introduction

• On Monday, we finished talking about framing by going over clocked based framing and SONET.
• We then talked about detecting bit errors using 2D parity, checksums, or CRC checks.
• We then started talking about reliable transmissions of frames.
• We went over the stop and wait protocol and the sliding window protocol.
• We said in the sliding window protocol, the sender keeps track of three numbers: a sender window size (SWS), a last acknowledgment received (LAR), and a last frame sent (LFS). We require `LFS - LAR \leq SWS.`
• The sender has timers for each unacknowledged frame and resends a frame whose timer goes off. When an acknowledgment is received, it is assumed all smaller values in the window are being acknowledged, so LAR is adjusted accordingly.
• Similarly, in the sliding window protocol, the receiver maintains: receiver window size (RWS), largest acceptable frame (LAF), and a last frame received (LFR). We require `LAF - LFR leq RWS`.
• When a receiver receives a frame, if all the frames between it and LFR have been received, then the largest sequence number not yet acknowledged is acknowledged.

More Sliding Window

• Suppose LFR=5, and RWS =4. So LAF = 9.
• If frames 7 and 8 are received, they will be buffered because they are within the receiver window. However, no ACK needs to be sent yet because frame 6 has not yet arrived.
• Frame 7 and 8 are said to be out of order.
• When frame 6 arrives (it might have come later because it had a problem and had to be resent), LFR is set to 8 (this is also the sequence number ACK'd) and LAF is set to 12.
• Suppose frame 6 had been lost, then a timeout would occur. The amount of data in transit decreases because the sender cannot advance his window. So in this situation the pipe might no longer be full.
• One way to mitigate against this problem is when frame 7 arrives, the receiver sends a negative acknowledgement (NAK) for frame 6. This might get back to the sender before its timeout expires and so would allow the sender to re-send more quickly. This, of course, adds complexity to the receiver.
• Another technique is to use selective acknowledgments. The receiver could acknowledge exactly those frames it receives, rather than the highest number frame received in order. This add considerable complexity though to the implementation.
• The sending window size, SWS, is selected according to how many frames we want to have outstanding on the link at a given time. It is set to how many frames fit in a given delay * bandwidth product.
• The RWS can be set to whatever the receiver wants. When RWS=1 the receiver does not buffer any out of order frames. When RWS=SWS, the receiver can buffer any of the frames the sender transmits. Setting RWS > SWS doesn't help since its impossible for more than SWS frames to arrive out of order.

Finite Sequence Numbers

• Let's assume now that sequence numbers are bounded to values less than some number MaxSeqNum.
• If the sequence number gets to MaxSeqNum - 1, the next sequence number is 0.
• How big does MaxSeqNum have to be so that we don't have to worry about two different incarnations of the same sequence number?
• Obviously, we want SWS < MaxSeqNum.
• If RWS=1, then MaxSeqNum = SWS + 1 will work.
• If SWS=RWS, SWS+1 does not work. Consider SWS=RWS=7, then if the sender sends 0..6 which are received by the receiver but all the ACKs are lost. Then the receiver is expecting frames 7,0..5. The sender timesout and resends 0..6 are some of these are confused for new frames.
• It turns out in this case the correct answers is SWS < (MaxSeqNum +1 )/2

In-Class Exercise

• Does anything go wrong in the last example if MaxSeqNum = 14? If not, 13? ...
• Explain your answer.
• Then post your solution to the Feb 15 In-class Exercise Thread.

Frame Order and Flow Control

The sliding window protocol is one of the best-known computer networking algorithms. It actually can be used to support three different roles:

• To deliver frames reliably across an unreliable link.
• To preserve the order in which frame are transmitted.
• To make sure that the receiver can slow down a sender who is transmitting too fast. (flow control)

Concurrent Logical Channels

• The data link layer in the ARPANET used an interesting alternative to the sliding window protocol.
• This protocol did not address order preservation or flow control.
• ARPANET supported the ability to multiplex several logical channels onto a single point-to-point link and then ran stop-and-wait for each of these channels. This idea was referred to as the support for concurrent logical channels.
• ARPANET supported 8 logical channels over each ground link and 16 over satellite links. Each channel had a 3 bit state, a boolean for whether the channel was busy, a 1-bit sequence number to use for the next frame, and the next sequence number to expect.
• When one wanted to send a frame one used the first non-busy channel.

Ethernet

• Ethernet is the most popular local area network (LAN) technology.
• It was developed at Xerox PARC in the mid-1970s.
• It makes use of a LAN technology known as carrier sense, multiple access with collision detection (CSMA/CD)
• Multiple nodes can send and receive data over a shared Ethernet link. This is what multiple access means.
• Carrier sense means that all nodes can distinguish between an idle link and if someone is talking (a busy link).
• Collision detection means that a node listens as it transmits and can detect when a frame it is transmitting is interfered with by another frame from another node.
• The technology that Ethernet uses was original worked out for an early packet radio network called Aloha developed at the University of Hawaii.
• Both systems had to solve the problem of how to use a shared resource fairly and efficiently. i.e., we need to figure out who can transmit when.
• DEC, Intel, and Xerox defined a 10 Mbps Ethernet standard in 1978, and this formed the basis for the IEEE 802.3 standard. This standard was later extended to include 100Mbps (Fast) Ethernet and 1000Mbps (Gigabit) Ethernet.
• We will over the next class describe how Ethernet works.

Physical Properties of Ethernet

• Originally, an Ethernet segment was implemented on a 50 ohm coaxial cable of up to 500m.
• Hosts connected to such a segment by tapping into it; taps had to be at least 2.5m apart.
• A transceiver, which was directly attached to the tap, detected when the line is idle and drove the signal when the host is transmitting.
• The transceiver also received incoming signals.
• It is connected to the Ethernet Adaptor of the given host.
• Multiple segments could be joined by repeaters up to a maximum total reach of 2500m.
• Bits on the Ethernet use Manchester encoding.
• Signals on Ethernet are heard by all hosts attached to it.
• Up 1,024 host may be attached to an Ethernet.

Variations on Physical Properties of Ethernet

• More recent Ethernets replace the 50 ohm cable (also called 10Base 5 or thicknet cable) with a thinner cable known as 10Base2. Here 10 means 10Mbps, and 2 refers to 200m rather than 500m.
• There is also 10BaseT Ethernet which uses twisted-pair wire and is restricted to 100m. This is what's used in Fast (100Mbps) Ethernet and in Gigabit Ethernet.
• The typical configuration for 10Base2 and 10BaseT is to have several point-to-point segments coming out of a multiway repeater rather than use taps. This kind of repeater is called a hub.
• Regardless of the set-up, all hosts can hear whatever anyone transmits on an Ethernet. So everyone on an Ethernet is competing for the shared resource. i.e., everyone is in the same collision domain.

Access Protocol

• We will look in a moment at the algorithm used by Ethernet to share the link. This algorithm is known as the media access control (MAC) algorithm.
• Before we do this, let us briefly refresh our knowledge about what an Ethernet frame looks like and what an Ethernet address looks like...

Ethernet Frame Format

• Ethernet is a bit-oriented framing protocol.
• The Preamble allows the receiver to synchronize with the signal; it is a sequence of alternating 0s and 1s.
• Both the Source and Destination hosts are identified by a 6 byte number. (We'll talk more about this on the next slide).
• The high-order bit of the destination address in IEEE 802.3 can be used to say single or group addresses (multicast).
• The Type in Digital-Intel-Xerox Ethernet is used to indicate to which of the many possible higher level protocols this frame should be delivered. In 802.3 Ethernet it is replaced with a 16-bit length field, and the first 16 bits of the Body contain the higher level protocol info. Since all type values are greater than 1500, it is possible using ANDing to get the two frame formats to coexist.
• The Body of the frame is at least 46 bytes long (so the whole frame is at least 64 bytes long). A whole frame can be at most 1500 bytes long.
• The CRC, of course, is used to detect errors.

Ethernet Addresses

• The 6 byte Ethernet address also called (MAC) is unique to each Network Adaptor (burned into its ROM).
• The IEEE manages a registry of the addresses (assigning different manufacturers different prefixes) which have been used so far. For example, the prefix 8:0:20 is for AMD.
• Addresses are usually written in hexadecimal with colons used to separate each byte and leading 0's dropped. For example: 8:0:2b:e4:b1:2 .
• Recall each digit in hex corresponds to a 4 bit nibble of the byte. So 2b is 0010 1011.
• FF:FF:FF:FF:FF:FF is used to indicate a broadcast frame for all hosts on the Ethernet.
• Each device on an Ethernet sees all frames that are sent on the Ethernet, but should only accept: (a) those addressed to it, (b) those broadcast to all hosts, or (c) those multi-cast to it. In the case where it is in promiscuous mode it should accept all frames.

MAC Algorithm

The smarts of Ethernet resides in how the sender sends frames it needs to send. It uses the following transmitter algorithm:

• If a network adaptor has a frame to send, and the line is idle, then the adaptor sends the frame immediately.
• If a network adaptor has a frame to send, and the line is busy, the adaptor waits for the line to become idle and then immediately send the frame. This is called 1-persistence; p-persistence would be to transmit with probability p and wait for the next frame "time slot" to consider transmitting with probability 1-p.
• If two transmitters try to transmit at the same time, then the frames are said to collide. The time it takes for a signal to transmit the whole Ethernet line length determines how many bits someone might have transmitted before noticing a collision. Once a collision is noticed, the transmitter is supposed to transmit a special 32-bit jamming sequence. This jammed frame is called a runt frame and contains at least the preamble and the jamming sequence (96 bits), and at most 512 bytes.
• Once the adaptor has detected a collision and stopped its transmission, it chooses to wait either 0s or 51.2μs at random; after which it starts transmitting again. If it fails again, the adaptor flips a four-sided coin and depending on its value tries again in either 0s, 51.2μs, 102.4μs, 153.6μs. If the kth try fails it flips a 2^k sided coin to get a number r between 0 and 2^k-1 and tries to transmit at time 51.2*r. An adaptor tries this up to k=16 before failing with an error. This is called exponential backoff.

In fast ethernet pseudo-random number generation is done using Linear feedback shift registers.

The worst case for detecting a collision is when the hosts are on opposite ends of the Ethernet. If d is the time for a signal to go the complete length of the Ethernet. Then Host A might transmit for d seconds and Host B might just start transmitting and immediately send a jamming sequence, this takes d second to get back to A. So A might transmit for 2*d before detecting a collision. On a 10Mbps Ethernet limited to 2500m, the delay would be at most 51.2μs and this would correspond to 512 bits = 96 bytes, which is why this is used as the shortest frame length.

Real world experience with Ethernet

• In the real-world, it has been observed that Ethernets tend to work best when they are lightly loaded.
• Utilization over 30% is considered heavy on Ethernet -- too much traffic is wasted on collisions.
• Since typically Ethernet have less than 200 hosts, and are far shorter than 2500m, both the round-trip delay and the likelihood of collision are much smaller.
• Ethernet is successful because it is reasonably easy to implement and administer. There is no hardware to maintain such as switches, cable is cheap, and the only cost to adding an additional node is the cost of a network adaptor.