Let us consider the specifics of different types of routing in industrial IP-network.
Broadcast
The broadcast peculiarity is the following: broadcast packets (data) are sent to all network devices without any exception. This type of addressing is initially considered as a method of service information exchange for devices operations coordination in the network. In this case, the logical question arises: why not to use broadcast for transmitting of voice information as well? This gives us the possibility to transmit sound messages to several subscriber devices at the same time. It seems very convenient but, in fact, there are a lot of side effects:
1. The enterprise network contains not only industrial communication system devices but servers, telephony, computers, printers as well. It means that the voice traffic will be sent to all these devices, the processors of which will have to handle this information. It results in parasitic load to the network.
2. In large companies the network is usually divided into segments (for example, several offices or remote departments). The broadcast is always limited with one network segment. This causes the problem of voice traffic transmitting between the network segments. The problem can be solved by using additional equipment (gateways) which converts the data into unicast format and forwards it to the gateway of the necessary segment and then the receiving gateway “unpacks” the information from unicast into broadcast. This procedure is called tunneling and is used for transmitting of broadcast packages to different segments of the network.
3. The state-of-the-art industrial communication systems often require integration with Wi-Fi, mobile networks and others which are communication channels with limited traffic capacity. When there are a lot of subscribers making calls, a great number of sent broadcast packets increase the load to the network which overloads slow channels. There can be a moment when there won’t be enough traffic capacity of Wi-Fi or of the mobile network for subscribers to make calls by themselves.
For example, the traffic capacity of DSL is 2,5 Mbit and one broadcast stream requires 512 Kbit. It means that 4 streams occupies one channel, so when it is necessary to create a large network of 100-200 subscribers, it is impossible to use DSL.
Speaking of the integration with Wi-Fi networks, their traffic capacity is nominally higher – 100 Mbit. However, the stated number is the peak value which can be reached by using two transmitters simultaneously or if the traffic is compressed. If the load to the Wi-Fi channel is slightly higher, the packets become damaged and instead of the continuous voice stream the subscriber gets intermission of the audio signal.
The other disadvantage is that Wi-Fi access points process broadcast packages incorrectly. It is connected with the fact that Wi-Fi points sent reports on delivery/ non-delivery of each packet. Thus, the traffic doubles for each sent packet and multiple deliveries of confirmation reports cause the considerable delay in access point operation. However, integration of Wi-Fi networks with IP solutions based on broadcast is possible but it will take additional efforts, namely: creation of the double dedicated network (for redundancy and increasing of the traffic capacity) especially for industrial communication system which brings all the distributed lines advantages to naught.
Multicast
Advantages of multicast routing come out of broadcast disadvantages. When using multicast the installer carries out the addressing setting by itself. It means that the user is able to specify which device will receive the voice information from other devices. It excludes the excessive load to all devices including ones with the limited traffic capacity.
Moreover, in state-of-the-art networks of the second generation the distribution of multicast packets is carried out by a switch and it is possible to transmit packages freely between different segments of the network (without using additional equipment). As a result, each device receives only traffic which is addressed to it and it is not engaged in rejection of unsolicited packets.
Thus, the user avoids parasitic load to the network and a part of intellectual load of the connections switching is moved to the computation capacity of the network. It results in the distributed system of industrial communication without the single point of failure and reimbursement from the existing IP-infrastructure.
See also:
Armtel distributed IPN system
Armtel centralized DCN system
Centralized vs Distributed Industrial Communications
