To answer the question, of course there is control, and it comes in the form of CSMA/CD (Carrier Sense Multiple Access / Collision Detection), a part of the IEEE 802.3 standard covering both baseband and broadband networks. IEEE 802.3 also covers a broad range of physical network cabling.

In this article I intend to cover the CSMA/CD process and how it works in the networks of today.

The Theory Behind CSMA/CD

To understand the theory of CSMA/CD, imagine you are with a group of friends and having a discussion. Each of your friends could be classified as a device on the network. At any one point in time only one person will be speaking (transmitting), and others will be listening (receiving). When one person finishes talking, the other people will see a blank spot in the speech and attempt to speak, and so keeping the conversation going.

If two people attempt to talk at the same time (collision), then both will stop, and then will wait for a random period (backoff algorithm) of time before speaking again. This will ensure that both get to say what they want to say, but one at a time and with no overlapping of what they are trying to say.

Throughout this article I will refer back to this conversation example, and how the IEEE 802.3 Ethernet standard can be applied back to it.

The OSI Layers

Because Ethernet is the cabling of the network, it works at layer 1 or the Physical Layer. This means that any protocol, whether it be TCP/IP, NetBeui, IPX or whatever can be used. Ethernet can however package the data differently before sending it onto the network, depending on the preference on the network and how the data is intended upon being used. If a frame is transmitted using one packaging type it cannot be read by a device using a different packaging type, so if you’re designing a network you’ll need to ensure you use the same packaging type on all devices. The most common types of Ethernet packaging are Ethernet II, 802.3 Ethernet, 802.2 SNAP (Subnetwork Architecture Protocol) and 802.2 SAP (Service Access Point).

With the conversation, you could look at this as having the actual message of the speech being in the higher layer protocols. So that by the time it is actually spoken (or in the case of Ethernet transmitted), it is the same no matter what method (or protocol), tone etc, is used when speaking. However it can be packaged differently, which we can look at as being different languages, where one person cannot understand more than one language.

Time Based Transmissions

Now we’ll have a look at the maths behind it all (unfortunately, yes you will need to apply some of that stuff from high school). All Ethernet MAC (Media Access Control) time-based transmission functions are measured by the amount of time it takes to transmit a single bit of data, this is more commonly called a bit-period. As the operating rate of the MAC increases (ie. 10Mbps to 100Mbps) the bit period reduces relative to real time.

With standard 10Mbps Ethernet operating at a bandwidth of 10Mhz, the bit period is 100 nanoseconds. Relative to this, if you increase the bandwidth to 100Mhz the bit period is reduced to 10 nanoseconds, and 1Ghz (1000Mhz) has the bit period set at 1 nanosecond.

In our conversation, a bit-period would be the amount of time taken to say a single word of data, with the entire sentence being the data frame.

This time based transmissions theory works well until such time as two devices decide to transmit at the same time, in which case a collision occurs. To handle this the 802.3 MAC has three other “time periods” for allowing devices to regain access to the transmission medium. These are the Slot-Time, InterFrame Gap and the Jam Period.

Slot-Time

As you know (well hopefully know anyway), Ethernet works by first listening for data to ensure that no other transmissions are...

Subscribe to our Free Must Know News Newsletter  Name:     Email: