To correctly build PON tree, it is necessary primarily to take into account, optical losses, brought by passive equipment. Theoretically, PON may cover area with radius 20 km. On practice - everything depends on losses budget on particular tree branch. It is necessary to be ruled when calculation by the worst attenuation, sensibility and radiation power rates of sensors.
When calculations attenuation tables 1 and 2 from section 2.3 are using for dividers (at that it is necessary to remember, that typical attenuation rates on different splitters are specified in tables without including attenuations on connectors).
On each welding optical signal attenuation takes as equal to 0.05 dB (good welding sets guarantee attenuation on welding 0.01dB, but it is better to be safe).
Attenuation on mechanical connection takes as equal to 0.5dB (in reality, the calibrated value for connection SC/UPC type is 0.25dB, but again, it is better to be safe - large optical budget storage never hurts anybody.
For the standard fiber G.652D the attenuation in fiber on waive length 1310nm is taken as equal to 0.36dB/km (for 1550nm the attenuation in fiber is equal to 0.22dB/km), the attenuation in kink is vary from 0.15dB to 7dB or more (this position is necessary to calculate by device at a place; when using external subscribers patch-cords with G.657 fiber. And it is possible not to take into account this position).
Further, it is necessary to "pass through" the already finished project, separating and summing that places, where there are elements, that add attenuation into the line. To add attenuation on fiber on all distance to received rate, from the most distanced subscriber to OLT. Thus, it is possible to calculate losses budget in PON.
The following step - optical budget of power calculation, but it is possible to take it as equal to 30dBm (manufacturer guarantees that the optical budget of power is equal to 30dBm). Everything that is higher than 30dBm is necessary to be tested.
If losses budget is lower than budget of power - the tree will be efficient, otherwise problems are possible.
*It is necessary to specify separately, that when PON calculating, where CATV and welded dividers are used, in is necessary to take into account, that TV signal has power from 7дBm (formed by cheap models of low power)to 24дBm (expensive transmitters or EDFA boosters), and this signal should be received by TV receiver (separate or included to ONU) with minimal power -12дBm(depends on equipment, but most often specifically -12dBm). That is why it is necessary to be very careful when using welded dividers and recalculate PON tree each time before new connection, to be sure, that each subscriber has same signal and is not over TV receiver sensibility.*
If when ONU installing (or second switching on/reloading from subscribers side) it couldn't register in the network - there is a high possibility that subscribers waive guide is damaged and signal attenuates higher than calculated (and of course, higher than PON optical budget). Damage may be connected with welding of low quality or with accidental kink on dividing node, as well as with lawbreakers actions or with other negative factors (tree branch is broken and fall on the cable, a lorry passed and broke patch-cord, etc.).
To solve this problems, it is necessary, first, to check all ONU condition on subscribers divider, where the problem ONU is connecting. ONU lets produce signal level monitoring at its receiver, as well as its transmitter power, sending these data to OLT at request. If all subscribers have this problem - to move in direction of parents tree node, widen the searching radius and checking on which level the problem is existing.
As soon as divider, that has some problem outcome, is found, ant other dividers are usable, be sure, you solved the problem. With a probability of 95% this problem is in the branch, that connects two nodes: parental half-working and daughterly non-working. It is quite easy to "light" this line by reflectometer, if PON is built on connectors, and extremely uneasy if the tree is "welded" (we cut the cable, weld the connector on, and than repair all).
Before actions, mentioned above, it is necessary to check ONU's laser transceiver responsibility ( anything can happen!)
There are situations, when ONU is destroyed and permanently radiates into the passive network on the waive length 1310nm (this situations happens extremely seldom), or unfair user connected to the tree instead of its ONU a powerful media converter, that radiates on the same waive length (1310nm). In these cases (as in case of hardware malfunction, as in case of sabotage) effect is the same: radiation presents permanently in riser tree flow and doesn't let other ONU to deliver data to OLT and "stops" OLT receiver by noise-type signal. As a result - the tree is not functioning.
There are two ways to solve the problem. The first one — to visit personally each house with a hope to find a breach or to find miraculously the law breacher. This variant have to be fulfilled, if PON is built by welding method of everything, that looks like fiber. If there are 20 subscribers, problem may be solved in weeks (or subscribers are not at home, or they don't want to open doors, or the weather is bad...).
The Second method is used when PON with the using of connectors is built – radiation localization by sequential tree branches switching off from «root» to «leafs». Steps consequence is as following:
- To switch off effects on root divider (and to check to be sure radiation presence), that go "down" across the tree; -When the branch with leakage radiation if found, to connect it off from the tree (the rest tree becomes usable) and to move down across this branch to the next divider; - To repeat until the problem is localized to the last (the farthest/the lowest in hierarchy) divider - then everything is clear. We switch consequentially off subscribers connection on the present divider, until we find radiation source.
Sometimes happens, that the cable is cut by competitors - this competition method is always popular for getting clients. But most often (especially in distanced districts) the cable is cut by locals by way of profit.
To avoid cable property invasions, it is necessary to explain to local treasure-hunters, that "There are no precious or other rare metals in this black cable, and its content is not useful even to tie up tomatoes" before cable installing in residential area. If it is possible, to demonstrate cable's part ans its content to locals. It doesn't let to avoid reasonable violation, but will take the wind out of sails of some sidecutters owners quantity.
Flood is the problem of each town network. Most often devices with burned port are flooding, rarely - users computers, captured by viruses. Also, users, that express some social protest (so-called) may flood reasonably, or that have some profits from inactive network. The main flood problem in FTTH is impossibility of remote access to flooding equipment or to localize programmatic the source.
Burned ports is mainly the problem of town type and is partially related to PON. As you know, as longer the copper conductor, as much magnetic fields are influencing on it. PON three is built on optic fiber, and, as a result, is not under influence of thunder. The problem may arise only in one case, if to one ONU several subscribers are connected by copper, and they are distanced from each other.
Everything is also clear with viruses and flooders: the first are using democratic freedoms, the second neglect all robotechnics rules. Method of controlling as first, as second are well known, but it is necessary to localize corrupting activity.
Everything is easy in PON. As it was mentioned above, PON is a centralizes managing system. All flows from subscribers come to OLT only. OLT only may give permission to ONU for data transferring, and OLT only may forbid any ONU contacts with external world. Active equipment in the PON tree absence, makes fight against flood much more easier - there are no cases with endless data packets resending with incorrect control sums between two neighbor commutative nodes, there are no buffers overrun there.
OLT is always connected to some senior device (for example, L3 router), that is why there is always asset to it. The senior device is not influenced by flood from clients in PON tree, because OLT separates specified time slot to each ONU and permit/forbid it a broadcast, so ONU can't uncontrolled "flood" OLT port and senior device with data packets. The same situation is with endless "walking" in the network traffic: there will be simply no of it, because each ONU has access to its neighbors-clients only through OLT.
Network monitoring process (for logical detection of flood source) may be organized with a help of optical port mirroring (where the whole tree "hangs") into copper and copper port connecting to some PC, where special soft is installed ( TCPDUMP for example).
Moreover, to protect network from virus activity, where there is uncontrolled MAC-addresses multiplication, there is a function, that forbids to separate (or to all) ONU to have more than N active MAC-addresses at the same time.