Professional Documents
Culture Documents
Bottom of Form
Note: Many topics at this site are reduced versions of the text in "The Encyclopedia of Networking and Telecommunications." Search results will not be as extensive as a search of the book's CD-ROM.
A token bus network is similar to a token ring network in that a station must have possession of a token before it can transmit on the network (see "Token and TokenPassing Access Methods"). However, the topology and token-passing method are different. The IEEE 802.4 Committee has defined token bus standards as broadband networks, as opposed to Ethernet's baseband transmission technique. The Manufacturing Automation Protocol (MAP) developed by General Motors for manufacturing floor networks uses IEEE 802.4. ARCNET (Attached Resource Computing Network) is a token bus network, although it does not conform to the IEEE 802.4 standards. The topology of the network can include groups of workstations connected by long trunk cables. These workstations branch from hubs in a star configuration, so the network has both a bus and star topology. Token bus topology is well suited to groups of users that are separated by some distance. IEEE 802.4 token bus networks are constructed with 75-ohm coaxial cable using a bus topology. The broadband characteristics of the 802.4 standard support transmission over several different channels simultaneously. The token and frames of data are passed from one station to another following the numeric sequence of the station addresses. Thus, the token follows a logical ring rather than a physical ring. The last station in numeric order passes the token back to the first station. The token does not follow the physical ordering of workstation attachment to the cable. Station 1 might be at one end of the cable and station 2 might be at the other, with station 3 in the middle. While token bus is used in some manufacturing environments, Ethernet and token ring standards have become more prominent in the office environment.
Number 3A
bandwidth management . Not designed to be the final implementation which would be a simpler protocol based on this protocol .
The Retrospective Token Bus Protocol can be used for all types of networks from office and telephone networks to industrial and home networks . It is ideally suited for use in Domestic Area Networks because of it's following advantages :1. low cost implementation - thus providing for a commercially attractive addition to domestic devices . 2. small IC real estate - thus allowing incorporation onto single chip solutions . 3. bus topology - low cost physical medium - can use existing cabling . Also highly reliable . 4. highly efficient - the low data link overhead means that the usage of the data link speed is maximised . 5. reliability and predictability of the time usage of the communication cycle - no drop outs due to packet collisions - ensures that intensive use such as pumping Digital Video around the home is conducted reliably - without drops outs and within the time requirements . The Retrospective Token Bus Protocol is ideally suited to Quality Of Service issues :1. it has the features and it can be vaired such that all QOS issues can be handled fully . 2. it is ideally suited as the LAN end of an ATM style network . It can also be used as a general LAN protocol . Please Note :- the information supplied here is only what is in the public domain . Commercially confidential information is not shown . This includes varations on the protocol that have not yet been patented .
Comparisons The Retrospective Token Bus Protocol versus the Ethernet Protocol
The Ethernet Protocol in a multidropped \ bussed topology ( eg. 10Base2 ) has a large and unpredictable data link overhead due to it's Carrier Sense Multiple Access \ Collision Detection access method . In large networks this can be easily 25% to 50% . The more units that are trying to transfer data at the same time the more the Ethernet crashes - the more drop outs and retries - hence the massive increase in data link overhead . There is also massive data link overhead with units waiting for the network to become available - inactive wait times . The Retrospective Token Bus Protocol , however , has a very low data link overhead of , typically , 0.25% to 0.5% and this overhead is very predictable . As such the Retrospective Token Bus Protocol is very suited to high network usage applications such as multi media networks such as domestic networks . The Ethernet Protocol in a star topology ( eg. 10BaseT - ie. requires a hub ) using packet switching does not have the data link overhead of the bussed topology . However the cost of the network is considerably greater than a bus topology based network . The Retrospective Token Bus Protocol , however , has low overhead and low cost .
The Retrospective Token Bus Protocol versus the Token Ring Protocol
The Token Ring network has high unreliability . If , for example , the network has 100 units and each unit's network interface has a reliability of 100,000 hours ( 12 years ) then the reliability of the network is 100,000 / 100 hours -> 1000 hours ( 6 weeks ) . This inherrent problem has to be designed around . Secondly the Token Ring network requires more wires - usually 6 - as opposed to 2 for a bussed topology . The Retrospective Token Bus Protocol , however , has the very high reliability and low cost of a bus network .
Wireless LAN's use CSMA\CD type access protocols and are similar in that respect to Ethernet . They have , as such , a high data link overhead . Their bandwidth is , also , low . Where a cable system can easily operate at 100 Mbs a wireless LAN will typically operate at only 1.6 Mbs - far too low for multimedia applications . Further , wireless LAN's have space and hence bandwidth contention problems in densely populated areas such as in cities . All these problems produce both a reliability problem and a performance problem for multimedia - ie. there is no guarantee that the multimedia packets will get through within sufficient time . Further there is the cost factor - a Retrospective Token Bus connection will cost typically U.S. 2.50 per unit whereas a Proxim LAN , for example , will cost approximately U.S. 129 .
Development
I would recommend that a version of Token Bus Protocol 2 be used with some Token Bus Protocol 3A techniques . This would use the active and passive token passing , arbitrary unit number assignment , network based addressing and broadcast channel addressing . It would essentially operate as a MAC style protocol - not requiring any MAC level intelligence . This would minimise the design complexity and the IC real estate required . Care would also have to be taken to ensure that the Data Link Level Media Access Control - works in all the practical situations - both protocol and physical wise .