1. Theoretical Capacity
The theoretical capacity of the MORSE system radio channel in the 25 kHz band at a modulation signalling rate of 21.68 kbps, after subtracting the effect of FEC (Forward Error Correction), is:
C = 21.68 : 1.375 = 15.76 kbps = approx. 16,000 bps = 2000 bytes/sec
i.e. the radio modem is capable of sending 2 bytes per 1ms.
This theoretical capacity is reduced by overheads which are necessary for transmitting each packet:
|time [ms]||equivalent [databyte]|
|synchronisation blocks||6 ms||12 byte|
|header||6 ms||12 byte|
|CRC + trailing bytes||4 ms||8 byte|
|RX/TX switching||2 ms||4 byte|
|ACK packet||15 ms||30 byte|
|total overheads||33 ms||66 byte|
|+ average access time to channel||37,5 ms||75 byte|
The algorithm or rather the access time to the channel is widely parametrisable. It is set up differently in collision networks and in master-slave networks.
The above-mentioned numbers indicate that the user data throughput in the MORSE system depends on the packet size. The optimum user data size per packet is somewhere between 200-400 bytes. For longer packets which are sent over a period of more than 300 ms there is a higher probability of repeating due to collisions or interference.
2. Non-Collision Network
The Master (control centre) calls one sub-station (PLC) after another, always waiting for delivery of a reply from the previous sub-station before calling the next sub-station in line. Such arrangement is called a Master-Slave system.
To calculate the period it is necessary to estimate the average number of retransmissions en route to the end station for the given network.
When calculating the time for calling all end points it is also necessary to calculate the time needed for processing replies and generating the next messages at the control centre.
Since the radio channel is not an ideal transmission medium the calculated theoretical response time needs to be increased by a reserve time for repeating under poor propagation conditions or repeating due to interference, etc. These and other unfavourable influences may increase the response time by up to 100 %.
A MS Excel program for calculating the period in a Master-Slave network is available to download here.
3. Collision Network
It is almost impossible to make an accurate theoretical calculation of the radio channel capacity in a collision network, because it is influenced by a large amount of parameters which are very difficult to define. We therefore make use of practical measuring and monitoring in order to determine the capacity of the radio channel in the MORSE system.
The setting of the algorithm for accessing the frequency of individual radio modems also influences the capacity of the radio channel: length and number of timeslots, ways of accessing the frequency (e.g. Reservation algorithm), number of repeats, method of acknowledgement (progressive ACK), etc.
From practical measurements we see that the typical capacity of the 25kHz radio channel in the MORSE system for full collision operation in one area of RF coverage is 600 bytes/sec.
600 bytes/sec corresponds to loading the channel to 30 % of its theoretical capacity, which is an above-average value in comparison with similar systems. In practice it is possible under certain conditions in the MORSE system to achieve up to 40 % loading, naturally at the expense of a disproportionate increase in the number of repeats and a marked loss of packets.
For 30 % loading the probability of repeating packets is in the range from 1:10 to 1:100, there are virtually no losses (better than 1:1000) and they generally correspond to the quality of the RF connection.