One of the biggest challenges in Planning, Designing and even Optimization of Mobile Networks is to identify where the users are, or how they are distributed.
Although this information is essential, it is not so easy to be obtained. But if we have and know how to use some counters related to this kind of analysis, everything is easier.
For GSM, we have seen that we can have a good idea of the location (distribution) of users through the measures of TA (Timing Advance), as we detailed in a tutorial about it.
Today we are going a little further, and know the equivalent parameters in other technologies, such as WCDMA (and LTE).
Learn the Performance Indicators related to the users distribution in a multi-technology mobile network, and also learn how to use these indicators together in analysis.
TA in 2G (GSM)
We’ve aready talked about TA in GSM in another tutorial, so let’s just remember the most important concept.
TA (Timing Advance) allows us to identify the distribution of 2G (GSM) users regarding its serving cell, based on signal propagation delay between the the UE’s and the BTS. The GSM mobile (from now on, we will call here UE too – as in 3G) receives data from BTS, and 3 time slots later sends its data. It is sufficient if the mobile is close to the BTS, however, when the UE is far away, it must take into account the delay that the signal will have to go through the radio path.
So: the UE sends the TA data together with other measures for the necessary time adjustments to be made.
In this way, we indirectly get a map with the distribution of users, or their probable location area, corresponding to the coverage area of the cell, with a minimum and maximum radius. The following figure shows this more clearly, for an antenna with 65 HBW, and maximum (1) and minimum (2) radius.
And in 3G and 4G (WCDMA, LTE), does we also have TA?
The expected question here is: does we have TA in 3G/4G? The answer is Yes, but in WCDMA the name is another, it is called Propagation Delay. (In LTE, we have both parameters – TA and PD).
So, let’s learn a little more about it.
Propagation Delay in 3G (WCDMA)
As we’ve told, in 3G the corresponding parameter to TA in 2G (GSM) is the Propagation Delay. With this parameter, we can estimate the distance between the UE and the serving cell, in the same way as we do in GSM.
But in 3G it has some different characteristics. To begin with, 3G measurements are made by the RNC, and not by the UE.
In one recent ‘RRC and RAB’ tutorial we have seen how an RRC connection is established, where the UE sends a ‘RRC CONNECTION MESSAGE’ message. When the RNC receives this message, it sends another message back to NodeB, to set up a Radio Link (‘RADIO LINK SETUP REQUEST’) (1). This message contains the Information Element with the Propagation Delay data, that is, the delay that has already been checked and adjusted to allow transmissions and reception synchronization.
As already mentioned, the information does not come from the UE as in GSM, but is the information that the RNC already has to make the communication possible: the information of this delay, the Propagation Delay Information Element (IE) is sent every 3 chips.
So let’s do some math.
- We know that the WCDMA has a constant rate equal to 3.84 Mcp chip/s.
- We also know (we consider) that the speed of light is 300,000 km/s.
In 1 second I have 3.84 M chips, in how many seconds I have 3 chips? Answer: 0.26 ps (pico seconds).
As we have seen that the information is sent every 3 chips, the total is 3 x 0.26 = 0.78 ps ps, which is the Propagation Delay time granularity.
And now let’s translate this minimum value into Distance: If I run 300,000 miles in 1 second, what distance I run in 0.78 ps? Answer: 234 meters.
In other words, have the Propagation Delay with granularity of 234 meters!
Note: it is important to know that this distance information is available to the system not only in the establishment of the call, but also during the entire existence of it.
Round Trip Delay – Round Trip Time (RTT)
When we talk about Propagation Delay, there’s another very important concept, related to the subject and used in several other areas that involve communication between two points: the Round Trip Delay & Time.
Let’s understand what it is with an example. Imagine a simple communication between two people, where the first say ‘Hi’, and the second one also answers ‘Hi’.
In an ideal world, first person speech travels up to the second one, taking a certain amount of time (t1), and the speech of the second person returns with a time (t2). So, we have a total time elapsed from when the first person said ‘hi’ till he received the other guy’s answer. This time is the Round Trip Time, or the time at which a signal travels a route until the response is received back at the source.
Bringing this analogy to an UE and a NodeB, we have the image below.
:: RTT = (t1 + t2)
In fact, the approach above is very close to real. But we have to consider also the time in which the receiver takes to ‘process’ the information, or the time it takes to respond after receiving the information.
Considering then this ‘latency’ time (TL), the RTT is so as:
:: RTT = (t1 + t2) + TL
So, we understand then what is RTT. But how do I use it?
This information is very important to the system, and can be used for several purposes. One of them for example, can be also to find UE’s locations. Our goal today is to know all means to find the location information of the UE’s, remember?
Well, this is another method (in addition to the counters, as we shall see soon). When the NodeB sends a message to the UE it knows exactly what time is. And then, when it receive a response from the UE, it also knows exactly that other time!
So, it just do the subtraction of the times to find the RTT, and calculate the distance! Note: the time used for the calculation is half of the RTT as the RTT is the round-trip path. In this case, the latency time on the receiver is ‘disregarded’.
With this distance information we can draw a circle with the likely area where the UE is. And if it is being served by various cells, the intersection of the circles of each one of them gives us a more accurate positioning (it is what we call ‘Triangulation’). And these calculations are even more accurate when other information is used togheter, such as ‘CellID’, MCC, RNC, LAC and Call Logs (CHR), with much more detailed information.
But let’s go back to the case where we only use the information of Propagation Delay – that is our focus today – and that already gives us sufficient allowance for several very interesting analysis.
TA and PD (Propagation Delay) counters
The Propagation Delay information are (also) available in simple form of Performance counters.
These types of counters are available in pre-set ranges according to each vendor. The ranges vary from 1 Propagation Delay to several ‘grouped’ Propagation Delay.
For example in Huawei have some TA ranges in GSM, and other PD ranges in WCDMA (Note: Huawei calls these propagation delay counter s as TP instead of PD). For an ‘ideal’ scenario, we would have counters for ‘each’ Propagation Delay.
Actually, that’s not what happens, because as we told before, they may be grouped into ranges. Note: the reason for this is not the case, but really too many ranges may even disrupt analysis.
TP (Propagation Delay WCDMA in Huawei) has 12 ranges.
In the above figure we have PDTA from 0 to 11.
- For TP_0 the UE is between 0 and 234 meters from NodeB;
- For TP_1 the UE is between 234 and 468 meters from NodeB;
- For TP_36_55 the UE is between 8.4 and 13.1 km from NodeB;
- And for TP_56_MORE the UE is more than 13.1 km from NodeB.
In the GSM (Huawei) have the same concept.
Note: See however that the amount of ranges here (GSM) is much bigger, and only begin to be grouped from 30 (from almost 17 km!).
With the counters organized in so different ways, be grouped by different ranges granularities, different distance (550 m for GSM and 234 m for WCDMA) it is very difficult to analyze the propagations, or rather, it is almost impossible to compare them…
And so what does we do, since we need to analyze the distribution of the UE’s in a generic way, doesn’t matter if it is using 2G or 3G?
The solution that we found in telecomHall was to make an ‘approach’, that is, a way to be able to see where we have more concentrated UE’s, no matter if at the time they are using 2G or 3G. Even because, this ‘distribution’ among Technologies and Carriers depends on several factors, such as selection and handover parameters, and also physical adjustments of radiant system. But the ‘concentration’ of users does not depend on these factors: the total amount of users in a particular area is always the same!
To this, the module ‘Hunter Propagation Analyzer’ uses a methodology and ‘particular’ counters, allowing to do this approach: we have created a range, and called it PDTA. As the 3G (Huawei, which we are using as an example) has less ranges – only 12, we made the initial PDTA definition based on it. The result can be seen in the table below.
Of course this approach or ‘methodology’ is not perfect, but in practice the outcome is very efficient. In addition, if you need a more detailed analysis (for example if you need to know with more accuracy than the approach presented here) just look to the original table, which contains each counter in its standard range in original granularity.
For other vendors, the ranges may be different, but the methodology is always the same.
In Ericsson for example, the Propagation Delay WCDMA counter is ‘pmPropagationDelay’, and it is collected by the RNC just like in Huawei.
It has 41 bins, being the first to indicate the maximum delay in chips (Cell Range), and other (1 to 40) to inform the number of samples in the period, referring to the percentage of the maximum Cell Range.
When the UE try to connect at one point greater than the Cell Range it will fail.
Regarding to bins, the distribution goes from 0 to 100%, as the rule below:
- bin1: samples between 0 and 1% of Cell Range (for example, if the Cell Range is 30 km, bin1 has the samples between 0 and 300 m from NodeB);
- bin2: samples between 1% and 2% of Cell Range;
- bin40: samples between 96% and 100% of Cell Range.
And the ‘adjust’ of PDTA can be done the same way, depending on your need.
Conclusion: Different vendors have different propagation counters, and in different formats – but the information is always the same! In all cases we can do the calculations that bring the analysis to the same comparison universe, with the benefits that we’ve illustrated above.
Distribution of Radio Link Failure (GSM) and EcNo (WCDMA)
Okay, we’ve seen today how to check the distribution of UE’s on 2G and/or 3G networks based on its counters. But in addition, we have also other equally interesting information!
In GSM, in addition to PDTA, we were able to count Radio Link Failures. And this gives us a great opportunity of crossing this information with the amount of Call Drops! The rule is simple: the point we have a lot of Radio Link Failures, ‘much’ probably we also have a lot of Dropped Calls! The relation is straightforward.
And in WCDMA, in addition to PDTA, we also have the average value of EcNo, that indicates the average quality of a given cell/region!
Note: In Huawei, for the average value of Ec/No for each TP, take the counter value and use the formula: EcNo = (value – 49) / 2.
TA in 4G (LTE)
As well as in 2G and 3G, we were also able to get the UE’s distribution information in LTE. The concepts applied are the same as already seen before, we can only point out that in LTE we have both TA and PD.
As today’s tutorial is already quite extensive, we will finish this part here, but with the certainty that if you assimilated what was presented, without any major problems you will be able to extend this information to your specific scenario.
After having seen – even with a little more detail – the concepts of propagation (including Failures in GSM and EcNo in WCDMA), we will see some possible analysis that we can do in practice.
We have already said that the professional who has experience on this kind of analysis can improve enough to network Indicators as Retainability and Accessibility. But how he manages to do this?
Simple: with the propagation analysis, it is possible to identify cells that are with their much greater coverage than planned/expected – ‘overshooting’ cells, especially if they are reaching places where we have other cells with better signal level!
In this case, we have pilot pollution, interference and high transmit power. As a result, increase of Establishment Failures and Call Drops, both in overshooting cell, as in the other where it is interfering.
In addition, we can discover cells that have their coverage area in the same direction (sector), but that have very different concentration (for example in the case of 3 WCDMA carrier, where one Carrier can be with the highest concentration of users closer to the cell, and another with this concentration away – don’t worry, we will see examples below and will be easier to understand).
This difference of distribution/concentration can be seen between the multi-technologies of the sector, for example, if the GSM coverage is much smaller than the WCDMA and vice versa. In this case, it serves as a great call for adjustments of tilts and azimuth between the antennas in this sector.
Practical analysis – Worksheets and Charts
Using data from simple counters, we already have excellent ways of analysis like charts and graphs. For example, the following is a complete view of a particular sector of our network (all cells of all technologies and all carriers). Note that the simple thematic distribution obtained with Excel Conditional Formatting already gives us a clear vision of this sector.
Filtering only for the contribution (‘PDTA_P’) of each cell, we can see clearly that a sector (Hxxx21) is with its coverage beyond the expected (1).
In addition, we were able to match (1) failures (now filtering by ‘ECNORLFAIL_P’), showing the immediate need for actions in this sector.
Practical Analysis – Maps
In addition to the simple analyses on charts and tables, we can geo-reference it, with a direct relationship with the coverage area. For demonstration, we create some dummy PDTA data of our network. Note: A real network has much more cells, but with these few sample data we can show the main points of analysis.
Continuing, we will then see the PDTA data of 4 examples sites plotted.
To analyze the PDTA distribution in Google Earth, we use a report generated by the ‘Hunter GE Propagation Analyzer’ module*, and so we need to know the criteria that we are using: in this report, the heights (1) from each region (PDTA of 0 to 11) represent the percentage of samples in that region. And the colors (2) represent the Quality: EcNo to UMTS, and Radio Link Failure % for GSM. *Note: you can build your reports in Google Earth and/or Mapinfo, just follow and apply the concepts presented here to your own tools/macros.
The data are grouped in ‘Folders’, with the first level being the sector (1) (a specific direction for all cells of all technologies and carriers). At the second level, we have the ranges (2) of PDTA percentage (how many samples from total cell samples we have in each region). And in the third level we have cells/PDTA (3).
Also equally important is the definition of the range used in the generation of the data, and consequently in the legend. Note that we use the same coloring scale for EcNo and Radio Link Failure. So, no matter if the coverage is GSM or UMTS – for example if the region is Red, we know it’s bad! (Or WCDMA EcNo worse than -16 dB, or GSM Radio Link Failure more than 50%!).
Knowing these details, we can do some demonstrations. Giving a zoom in a more extensive area, we see that we have multiple cells with coverage in places where they should not be covering. Of course, these points have a few samples, but with vary bad quality, as we see in the region shown below (1) – ranges mostly Pink, Red and Orange.
Analyzing specific cells, for example ‘AAN’, we see that the same coverage area is much larger than it should (overshooting cell), both the GSM (1) and UMTS (2) are more than 4 km of the serving cell.
In this case, we have another interesting point, also seen below: most of the users in the region (1) are served almost exclusively by GSM. Now in region (2) almost all users use WCDMA. This is another point of optimization: these coverages should be, as far as possible, ‘proportional’.
Another example: the ‘ABU’ site is a typical case of need of urgent action, for example by increasing the tilt’s of overshooting cells. Too many samples at more than 4 km, and with poor quality. As these are cells of an urban area, and in addition we have other cells serving that distant locations, it is recommended to increase tilt, and later run a new analysis.
The opposite of what we saw above is also possible: we can identify cells that have a very good coverage area (in this case, a more contained area), and with excellent quality levels (Green and Blue).
We could go on demonstrating several other analyses that are possible using the data presented here today. However, the best way is that you use these incredible resource in your analysis, because with no doubt it represents a big help.
Many people try to optimize the network based on parameter changes only. But we saw that in many cases like above, there may be situations where the most recommended is physical intervention (adjusting of Antenna, Height, Azimuth, Tilt, etc…).
No doubt the analysis presented in this tutorial are essential to the improvement of any mobile network, and if you so far haven’t used, it’s a good time to start.