Service Levels for Microwave Networks
Defining service levels for microwave networks is always an argument between good engineering and budget control. The allure of microwave is the speed and cost of deployment in comparison to a fibre network. What must not be lost is the fundamental reasons for building a network in the first place and that is to support the traffic requirements of the higher application.
There are 2 end to end performance considerations when designing networks:
- Availability – amount of time in a calendar year that the network can be down due to fault. This is usually expressed as a percentage of up time eg. 99.99% which correlates to around 53 minutes a year. The determination of unavailability is on the receipt of 10 severely errored seconds and it is considered cleared on the receipt of 10 non severely errored seconds of data.
- Error performance – short duration hits (less than 10 seconds) that result in errors to the data but are not classified as unavailability.
Availability is determined by designing in redundancy into critical systems and diversity in routes i.e. protection, to ensure that faults can be worked around until repaired. This is especially important in remote areas where engineers may require a number of hours to reach a site. Early detection of faults is critical in high availability networks requiring real time monitoring systems and well trained staff to interpret the status of the network, especially in times of instability due to seasonal fading.
Error performance is determined by designing each microwave hop to ensure that the cumulative performance end-to-end is within specification. The mistake often made is to design each hop to the desired error performance of the network. Error performance is improved by using suitable bands for the length of the hop, designing for a high fade margin, ensuring that Fresnel Zone clearance is available based on the length of the hop and the climate in that region, minimising multipath fading through techniques such as space diversity, frequency diversity, and avoidance of paths over water and other topographical features etc.
Error performance is usually specified as predicted outage in seconds worst month. Multipath fading is the most likely cause of poor error performance and the techniques for mitigation are costly:
- Fade margin – higher fade margin will usually be gained through larger antennas. These antennas are a significant cost but are nothing compared to the costs for strengthening of structures and the operational expenditure for leasing space on third party towers.
- Space Diversity – requires a second antenna on the receiver (usually both ends) doubling the load on the structure. The costs for the radio systems tend to double as a space diversity receiver is required ( IF combiner preferred to baseband switching) as well as the additional antenna, feeder etc. Indoor systems have the additional costs of waveguide and pressurisation systems. All in all this link is starting to look expensive from a capital budget perspective.
It is usually at this point when the capital costs are steeply rising not to mention the operational expenditure for the additional site costs that some compromises start to slip in. Structural upgrades are revisited and savings found by lowering antennas, reducing size of antennas, or removing space diversity or even reducing the strengthening work and accepting more movement in high winds. Once the compromise is made on one hop, suddenly all hops are reappraised for the same savings until the budget is brought under control. Each compromise had some affect on the error performance but the results are usually not known until much later during the commissioning phase or when the collection of operational data starts to show sub performing links.
The cost of the lower performance is transferred to the higher application data eg. Television pictures, telephony, public safety networks communications, etc. Remedial works are now required and it takes some time to see whether these have been successful. It is often not possible to go back and strengthen the structure and swap out the antennas for the correct configuration due to the significant outages required and so compromises are the order of the day. The end result is that the initial cost savings during construction are often surpassed by the cost to the application data and remedial works with the resultant improvement in error performance being less than desired.
Surely this would never happen I hear you say but sadly I have seen this on many significant networks.
The answer is to properly define service levels based on the higher application requirements. Once defined, engineering must be allowed to design and build to support those requirements otherwise any savings made during this phase will be paid for many times over the life of the network.
If you require an assessment of an existing network / links or a review of a design please contact us for a consultation.