Communications, Protocols and Polling on Distributed Satellite- Local Networks

This article is intended as areview of the general methods employed today with regards to communications for distributed networks. Discussion will center on the use of satellite networks and local networks.

The different types of ground-based communications available will be touched on lightly, with the use of examples from techniques currently being employed by Scientific-Atlanta whenever possible. Discussion of satellite-based communications technique will center mostly o current theory employed int the research community. Protocols that are discussed will include the usage of SAbus protocol for local network communications, and the use of variations of atheoretical type protocol for satellite networking.

A discussion of methods employed by Scientific-Atlanta for the polling devices in a local network will be presented, with examples, provided on the capabilities from a recent projects. In addition, satellite networking will be discussed with regard to the differences in methodology required in polling earth stations and sending commands.

Distributed local area networks can be organized in may different fashions. Some of the approaches usef for network configurations are:

Star network--Each device is connected separately to a central controlling point in the network.

Bus network (or multidrop)--Places all the devices in the network together on a single shared line.

Ring network (or loop)--Where each device is connected to each other in a closed circular fashion, providing two pathways around the ring, going in either direction. Breaking the ring at one spot would not affect the network's functionally.

Mesh network--Any combination of any of the above networks, with each connecting node representing either a single device or a local network comprised of a star, bus or ring design. Typically used over large geographical alreas.

Devices in a network configuration can be connected in either a point-to-point fashion, or they may be multiplexed at some point in the ntetwork (generally to remote sites.) Most often, multiplexing is done over long distances to reduce cost and/or complexity, by concentrating the many lines used for devices into one line. Multiplexers are either time-division (TDM) or frequently-division (FDM)--dependent techniques for splitting up the many incoming signals, over a single physical line.

Types of line connections made between devices are general made via one or a combination of the following: locally by means of a cable--either coaxial, twisted-wire pairs or fiber-optic; line-of-site by either microwave, or laser; ba the telephone company (telco) on either a regular phone line or on a dedicated leased-line connection.

As a commercial user's needs grow, and as networks of earth station sites evolve, the need arises for less dependence on the common carrier's (telco) lines of communication. More emphasis is placed on taking advantage of satellites not only for the distribution of video and auudio signals, but also for controlling the flow of signal traffic and using the satellite facilities as a means of interconnecting users of the network. By using the satellite as part of a shared distributed network, earth station affiliates may now obtain all of cost imposed by the telcos. Additionally, users can receive instructions on daily opperations via satellite as well. If the network user is a receive-only earth stations, then they must still rely on lanlines to repsond to other users in the network. However, transmit and receive earth stations may respond via satellite as well.

Some disadavantages of using satellites are that one can only cover a third of the earth's surface with a single satellite (in actuality this is generally less). No privacy is available in a satellite-based network without the use of data-encryption techniques. Satellite channels may be inadavertently jammed or blasted by maliciousness or ignorance at uplink earht stations. The channel can be saturated when peak traffic times occur. All of these problems occur in an equivalent fashion over terrestrial lines, so it should be recognized that they are not unique to satellites.

Some of the advantages of using satellite-based communications are that packetfor more than one user are boradcast once. New users can enter the network without need to rewire is posisible and it can be easily expanded. Mobile users are handled without troubles. Packet switching is handled without the need for packet switches.

A major component of any satellite-based network is the uplink earth sation facilities. These facilitieis serve as a hub for a local or regional area network. All receive-only stations will generally connect to the closest uplink via a ground line, so that they may respond to polling of status and commands if need be.

With regard to packets of data sent over the satellite network, an uplink site has the ability ot monitor satellite traffic for collisions on the transmissions it makes with other uplink sites. This would generally be avoided by a schmeme of roun-robin, discrete-time intervals, in which each uplink station would transmit commands or status of receive-only users in its local network. Collision avoidance can also be handled by putting different uplink transmissions on different channels on the same satellite, of by putting the transmission on different satellites altogether. Being albe to monitor other receive-only stations and control them with commands from the uplink allows greater control over the functioning of a network, and the routing of packets through the network.

Downlink facilities are earth stations equipped only with the ability to receive signals, be they video or audio. A facility must still use a ground link to communicate with other users in the satellite-based network. Receive-only earth stations are far less costly than uplink facilities and are therefore more affordable for the smaller networks.

The multidrop approach to local networking follows what is typically known as statistical multiplexing using centralized polling. The polling controller or master device on a multidrop bus will request status from each of the slave devices on the bus (twisted-wire pair calbe for an SAbus). Upon receiving a poll for status fromthe master device, a slave will respond within a set number of milliseconds with data describing briefly the slave device's current stae. If a state change has occurred since the last poll, then the master device can poll for addiational information. The master device then moves on to the next device on the bus.

Each time a poll is made, the slave device ahs to be identified via an address specification, which is unique for each slave device. Overhead required for a poll includes information on the start of text, an identifying slave device address, the data itself, an end-of/text identifier and a check-sum value. The data sent in a poll, and responded to by the slave device, generally consists of a command followed by any pertinent parameters needed. According to Rosnes, "The key point is that polling is a prime example of a system which uses dynamic, rather than fixed, assignment of the capacity of the transmission of the capacity is shared, follwing some formalized assignment procedures among the active members of the user community, and little or no capacity is wated by the users that are currently inactive. In the case of the multidrop bus, slave devices simpley listen to each poll message on the line and ignore any messages that are not addressed to itself.

Typical master devices can include the following: master control computer, which has the ability to centrally control all elements of remote and local earth staions; and can easily be adapted to control various equipment protocols, most commonly via an RS-232Cable; earth-station controller; video protection switch; and uplink switch.

Typical slave devices can include the following: video protection swith; LNA protecton swtich; uplink protection switch; receiver; exciter; antenna controller; klystron tuner; high-power amplifier; waveguide switch interface; and device connected to the bus via software and/or hardware drive interface.

A feasible approach to providing a satellite network protocol would be a slotted Aloha protocol technique. To describe briefly its approach, one should know that the following must be employed:

Establish a synchronized time interval over a satellite channel, according to the time to transmit the length of the largest packet message.

Broadcast a packet only at the beginning of thsi discrete-time interval, for example every 20 milliseconds.

Provide an round-robin polling approach, with the ability to reprogram the order of poll and the priority.

The packer itself would contain a hello/message/goodbye structure.

Collisions occasionally occur due to quareter-second (250 ms) delays in transmissions.

Uplinks would listen to their own messages and compaere with that was transmitted; if a a collision has occurred, then it would retry the transmission when its time slice comes up again.

"Packer narrowcasting; would be employed, which implies that the packet contains source and destination addresses, as well as a repetition counter for repeat transmits of the same packets.

"Capture effect" could also be used, where different uplinks would broadcast signals at different power levels, thereby providing a pseudo-priority basis for resolving collision conflicts. In other words, stronger signals to the satellite provide a higher priority by increasing the probability that its packet would get through. If the highest priority makes it through the satellite transponder successfully, then you have one less uplink packeto retransmit in the event of a collision.

In a local area network, polling performs a request for status in ar ound-robin fashion, over a twisted-wire pari cable. Each poll expects a response, within a specified time frame, form the device. In order that futuer requirements be accommodated, the current polling module being used for projects has been designed to allow for the polling of SAbuses (or, in fact, any type bus interfaceable through a DEC-compatible interface board, RS-232C, RS-422 and others) in any order, as well as poll the devices on a bus in any order. (figure 1.)

The frequency with which devices or buses ar polled can be handled in the same variable fashion. The method used was to place a bus numbers in a variable size array which can have the buses arranged in any order; in addition, one is able to repeat the same number in the array several times (effectively increasing the frequency of polling that particular bus.)

Therefore, the buses cna be prioritized and/or have the frequency of polling an individual bus increased at the expense of others. The same approach was applied to the polling of devices on each bus, providing the ability to vary both the order in which devices are polled and the frequency with which individual devices are polled.

Satellite-based network polling must be handled in a different fashion than ground-based networking, due to the nature of communications. Because of delays in transmission and the fact that not all earth stations have an uplink facility, a more applicable approach might be the use of a slotted Aloha protocol In polling the stations, a response would not be required if the master earth station or controller broadcast what it assumed to be the current state of each station, and then received a response from the slave stations only when the state chnaged. This change notice could be supplied via landlines for receive-only earth station, and via satellite for transmit-and-receive stations.

Commands over satellite networks can be handled in similar fashion, whereby a response is given from a problem exists in meeting the demands of the command. Additionally, commands can be issued which are either global (pertaining to all stations), group (pertaining to a subset, such as regional) or specific (pertaining to a single earth station). Commands could aslo be submitted to all stations hurs in advance of theri actual execution to ensure their proper execution at the proper time. This would also assist in cutting down on unnecessary traffic during peak times. In addition, one can prioritize traffice such that the most critical packets of commands get through fastest, with predence over polling packets and others low-priority commands.

To provide privacy to the transmission of polling and command data packets, encryption may be employed. It is possible to protect, via software encryption, both the status report of stations and the global, group or specific commands to the network. Data can be protected from outside eavesdroppers on the network, as well as providing protection for stations within the network itself. This is achieved by providing as many access keys and/or encryption methods as is necessary to keep outside, group or specific user eavesdropping from occurring. Electronic signatures can be changed on a frequent basis to provided protection against the accidental (or otherwise) disclosures of the access keys.

COPYRIGHT 1984 Nelson Publishing
COPYRIGHT 2004 Gale Group