Why increasing PS PSINACTTMRFORINT increase PS R99 drop rate ?

[fahmi]

Dear,

In our network

PSINACTTMRFORINT=900s
PSINACTTMRFORBAC=900s

and this obviously increase call drop rate (recommanded value =20s). This mean being so long in CELL PCH stae increase the call drop !

Why?

BR,
Fahmi

[gunpowder]

Dear,

By increasing PSInactivity timer you are decreasing attempts if you are applying fast dormancy on your network you are advised to change PS CDR to the EFD PS Drop Formula

[Lazaros]

True! Call Drops in Cell_PCH are not considered drops in EFD Huawei Formula. Consult Huawei Hedex and update it!

[samuraial]

If you change it to 10s you will further improve CDR and increase capacity :)

[rfsupp]

Fahmi, the old formula doesn't consider the transition to CELL_PCH as a normal release and also the vendor sometime consider that if you are on CELL_PCH you cant have drop because you dont have any RL established. So this drop increasing isnt user experienced

Samuraial, Increase capacity reducing this timer? You going to wrong way dude.
With a small value you have more RRC reattempts, so you increase RNC/Nodeb board load and reduce the cell resources.

[samuraial]

@rfsupp
Yeah your are right about RRC load and xpu load increasing in RNC but keeping a user long even if there is no data transmitted means he is using resources.
The longer PS inactiive timer, more radio resource occupation.
That's why there is a balance and 10s seems to be acceptable for most cases i have seen.

B/R

[rfsupp]

@samuraial,

Using resources? CELL_PCH doesn't have any RL link!!! So there is no resource occupation.

In CELL_PCH you have Idle mode behavior, but you dont need to perform RRC procedure again because you maintain U-RNTI.
And the difference between CELL_PCH and Idle Mode is that the first state needs to perform cell updates(uplink) every time that reselection is performed.

And if the user is in CELL_DCH, the state transition shall move such UE to CELL_FACH and than to CELL_PCH....So again no increase of resource consumption.

[samuraial]

@rfsup
PsInactTmrForInt
Meaning: When detecting that the Ps' Interactive User had no data to transfer for a long time which longer than this timer, the PDCP layer would request the RRC layer to release this Radio Access Bear.
This is the explanation from M2000.
I was refering to users staying in DCH about resource consumtion(specially CE usage.)
Different network, different strategies.

B/R

[rfsupp]

Even in this way you are forgetting to optimize the State Transition....Even if I define PsInactTmrForInt as 1800s. I wont impact the resources in CELL_DCH, if the state transition is ok.

Usually the state transition total timer is around 5s-8s, so you are saying PsInactTmrForInt is 10s. With this approach the UE will be moved to CELL_FACH at least if D2P is not activated.
So at this time the user is in common channel sharing the resources and more 5s-8s the UE is in CELL_PCH using no resources.

As the smartphone devices is a big percentage of the UE devices, just one UE may try to connect 100 times/s due to apps(Instagram,Facebook, twitter and etc). So in this way you can imagine how much this kind of UE will impact the Cell Consumption, in the other hand, PsInactTmrForInt 1800s I maintain the UE in CELL_PCH (not in CELL_DCH due to State Transition algorithm). Using just 4 messages to reconnect again instead 20 messages in case you send the UE to Idle Mode with PsInactTmrForInt 10s.

I agree that is different strategies, but spending much more CAPEX and higher PS setup delay.

[tharinduwije]

I think it should be a balance between very low & very long PsInactTmrForInt.if its too long resources will be wasted too much if its too short signalling load will be too much creating high load on NodeB & RNC.