VoLTE KPIs

Today we will discuss about the Various types of VoLTE KPIs. 
VoLTE KPI is mainly divided into three parts:

1.IMS KPI

2.EPC(Packet core) KPI

3.RAN KPI.


IMS KPI is further divided into 2 parts:- IMS Control plane KPI and IMS User Plan KPI.

IMS Control plane KPIs are as follows:

1.RSR:- Registration success rate.         

Formulae = Count of 200OK for Registration completed/Count of SIP register sent from UE excluding 401 error attempts*100

2.CSSR- Call setup success rate%           

Formulae = Count of (Normal end call+ call failed with user behaviour)/Sum of all call attempts*100

3.Call setup time:-          

Formulae= Avg of(Time(180 ringing)- Time(SIP INVITE request))


IMS User plane KPIs are as follows:

1.Mute rate(%)

Formulae= % of calls muted (samples >2 or 5s RTP loss in both direction)

2.MOS score

3.RTP packet loss %

Formulae = % of RTP packets lost in the uplink or downlink direction.

4.One way call %

Formulae = % of calls having no voice packets count for 2 or 5 Sec in either the downlink or uplink direction.


EPC(Packet core KPI):- This also classified further in 2 parts- MME generated and SGw/PGw generated.

MME generated KPI are as follows:-

1.VoLTE attach success rate% = % of PDN connect request responded successfully by MME for IMS APN.

2.VoLTE bearer Activation success rate:% of create request got successful.

3.Paging success rate%-= % of paging responses on QCI=5 received at MME.
SGw/PGw generated KPI are as follows:-
1. IMS IP Pool Utilization%


RAN KPI are as follows: 
1.Call drop rate % = % of calls getting dropped abnormally.
2.SRVCC Success Rate % = % of calls successfully transferred with SRVCC from VoLTE to legacy NW

3.Handover Success rate %(S1 and X2).

5G NR network Architecture

Below is the 5G NR architecture as per 3GPP standard.

Network nodes and their functions:

(UE)- User equipment

(gNB)- Next Generation Node Base station

Access and Mobility Management Function (AMF) – responsible for following

       – Termination of RAN Control Plane interface (NG2)

       – Termination of NAS (NG1), NAS ciphering and integrity protection

       – Mobility Management

       – Lawful intercept (for AMF events and interface to Lawful Intercept System)

       – Transparent proxy for routing access authentication and SM messages

       – Access Authentication

       – Access Authorization

       – Security Anchor Function (SEA): It interacts with the UDM and the UE, receives the       intermediate key that was established as a result of the UE authentication process; in case of USIM based authentication, the AMF retrieves the security material from the UDM

       – Security Context Management (SCM): it receives a key from the SEA that it uses to derive access-network specific keys

User plane Function (UPF) – functions are

       – QoS handling for User plane

       – Packet routing & forwarding

       – Packet inspection and Policy rule enforcement

       – Lawful intercept (User Plane)

       – Traffic accounting and reporting

       – Anchor point for Intra-/Inter-RAT mobility (when applicable)

       – Support for interaction with external DN for transport of signaling for PDU session authorization/authentication by external DN

Session Management Control Function (SMF) – supports following:

      – Session Management

      – UE IP address allocation & management (including optional Authorization).

      – Selection and control of User Plane function

      – Termination of interfaces towards Policy control and Charging functions

      – Control part of policy enforcement and QoS.

      – Lawful intercept (for Session Management events and interface to Lawful Intercept System)

      – Termination of Session Management parts of NAS messages

      – Downlink Data Notification

      – Initiator of Access Node specific Session Management information, sent via AMF over NG2 to Access Node

      – Roaming functionality

      – Handle local enforcement to apply QoS SLAs (VPLMN)

      – Charging data collection and charging interface (VPLMN)

      – Lawful intercept (in VPLMN for Session Management events and interface to Lawful Intercept System)

Data Network (DN): Operator services, Internet access or other services

Authentication Server Function (AUSF) – Performs authentication processes with the UE

Unified Data Management (UDM) – Supports:

      – Authentication Credential Repository and Processing Function (ARPF); this function stores the long-term security credentials used in authentication for AKA

      – Storing of Subscription information

Policy Control Function (PCF) – Provides:

     – Support of unified policy framework to govern network behavior

     – Policy rules to control plane function(s) that enforce them

Application Function (AF) – Requests dynamic policies and/or charging control

Rank Indicator(RI) in LTE

The basic advantage of MIMO(Multiple Input Multiple Output) technology is that in the MIMO technology different data streams are transmitted simultaneously on the same frequency and time.

https://mytechnew2019.blogspot.com/2019/01/benefits-of-mimo-technology.html

      Rank Indicator(RI) is an indicator which tells how well the multiple antenna works in the MIMO, i.e. the number of layers transmitted simultaneously on the same frequency and time.

Calculation of maximum RI:- Maximum RI is very close to number of antennas in Transmitting and receiving side. Suppose we have 2*2 MIMO( 2 Antennas in both transmitting and Receiving) then the value of RI are 1 and 2.

Note: If we have different number of  transmitting and receiving antennas then the maximum RI is equal to one of less antennas.

How RI works:- Lets take an example of 2*2 MIMO. The possibilities of RI are 1 and 2. 
                           The UE reports RI 2 when there is No Correlation(Interference) between both antennas i.e different data streams are being send by each TX(enhanced capacity).
                            The UE reports RI 1 when there is some interference between both antennas, i.e all the TX are sending same data streams(enhance coverage).

The feedback of RI by UE is based on the SINR value. For a UE to be able to support 2 layers (RI2) means that it has to be in a good SNR. The second thing in multipath rich environment, UEs prefer RI2 as it has embedded copies of the data streams. The third thing is the vendor's implementation. Some vendors switch on the basis of SINR only while other vendors switch on the basis of SINR plus FER while there are vendors that switch on the basis of SINR plus HARQ-Outs.

                        

Cell Reselection in LTE

Cell Reselection Procedure in LTE:- There are some procedure followed in cell reselection for LTE listed as below:-

1.Serving Cell Measurment
2.Cell Reselection Triggering
3. Cell Ranking 
4. Cell Reselection.
We will discuss one  by one.

1. Serving Cell Measurement:- UE, in Idle state, wakes up at the end of every DRX cycle to measure the signal of its serving cell (Qrxlevmeas) and calculate the received signal level (Srxlev) of the serving cell to decide whether it should stay or move to another cell.

2. Cell Reselection Triggering:- If the received signal level of the serving cell (Srxlev) is greater than the specified threshold value (s-IntraSearch for Intra Frequency and S-NonIntraSearch for Inter frequency)), the UE stays in the current serving cell. If not, it triggers a cell reselection procedure. The threshold values are delivered through SIB 3, and defined as s-IntraSearch in Release 8 and as s-IntraSearchP and s-IntraSearchQ in Release 9.



3. Cell Ranking:- The UE ranks each cell (Rs, Rn) based on the measured signal strength of the serving cell (Qmeas,s) and neighbor cells (Qmeas,n). Parameters required for cell ranking are delivered through SIBs 3 and 4. The serving cell is ranked using the hysteresis (q-Hyst) value stored in SIB 3 while the neighbor cells are ranked based on the offset (q-OffsetCell) value specified for each cell in SIB 4.

4.Cell Reselection:- After serving cell and neighbor cells are ranked, the UE decides whether the cell reselection criterion is satisfied (Rn > Rs) or not. If there are neighbor cell(s) that satisfy the criteria, the UE selects the best satisfying cell, and then reselct there. Cell reselection is performed only when the criterion is satisfied for a certain period of time (t-ReselectionEUTRA).



We can prevent too frequent cell reselection and make sure reselection is performed in proper manner by hysteresis and cell-specific offset values. In addition, we can control the q-Hyst and t-ReselectionEUTRA values by applying appropriate scaling factor (q-hystSF, t-ReselectionEUTRA-SF) depending on the traveling speed of the UE.

Below table is the parameter summary for the cell reselection.

Parametrs	Description	SIB Type
S-Intrasearch	Value that triggers Intra frequency measurment(dB)	3
q-Hyst	Hysteresis value for Serving cell ranking	3
q-RxlevMin	Minimum Rx Level required for UE to continue on serving Cell	3
p-Max	Maximum TX power allowed for UE	3
allowedMeasBandwidth	DL BW to be measured by UE	3
t-ReselectionEUtra	Cell Reselection timer Value	3
physcellID	PCI of neighbouring cell	4
q-offsetCell	offset value for serving cell ranking	4
intrafreqblackCelllist	List of neighbour cells that are black listed for reselection	4
physcellidrange	PCI Range	4

 

MRO Feature in LTE

MRO(Mobility Robustness Optimization):- MRO is a SON(Self Optimizing Network) feature in which they provide the solution for automatic detection and correction of errors(Radio Link Failure) in Handover.

Scenarios for MRO:- There are 3 scenarios where MRO feature can optimize in case of RLF in Handover.

1. Too Late Handover.
2. Too Early Handover.
3. Handover to Wrong Cell.

We will discuss one by one.

1. Too Late Handover:- In this case the UE does not receive the RRC Handover command, due to weak signal see figure below, the handover procedure in the source cell is initialized too late, since the UE is moving faster than the Handover (HO) parameter settings allow. Hence when the RRC HO command from the serving cell is transmitted the signal strength is too weak to reach the UE, which is now  located in the target cell, connection is lost. The UE attempts a connection re-establishment, containing PCID and C-RNTI belonging to the source cell, but received by the target cell. The target eNB will then inform the source cell about RLF to adjust Handover parameters.

2. Too Early Handover:-  In this scenario the signal strength in the target cell is too weak, and the connection is lost almost immediately after the Handover. see figure below. The UE has successfully been handed over from source cell to target cell , but since it was triggered too early the connection will drop almost immediately due to too poor radio conditions in the target cell . The UE will then re-established the connection in source cell.

3. Handover to Wrong Cell:- In this case RLF occurs in the target cell after a handover has been completed, and the UE attempts to re-establish its radio link in a cell which is not the source cell nor the target cell. See the fig below.

Working of MRO Feature:- Source eNodeB takes some appropriate actions to overcome the above type of failures by changing the threshold at which the handovers are triggered. Source eNodeB can use ‘Mobility Settings Change’ procedure to inform the neighbour eNodeB of the change of threshold it has performed. In this case the ‘MOBILITY CHANGE REQUEST’ message is sent with a cause value ‘handover optimization’ and indicates the change (in dB) of the handover trigger parameter change performed in the source cell. For instance, in the case of repeated handover to an

inappropriate cell, the source eNodeB could modify the way it builds its list of candidate

target cells.

The Parameters that can be optimized in connected mode: The majorly parameter modification are done for A3 events triggering as listed below.

1. A3 offset

2. A3 hysteresis.

3. Cell individual offset(CIO)

4. Time to trigger(TTT).

Ref: 3GPP.ORG

Carrier Aggregation in LTE

What is Aggregation?:- In Layman language the meaning of Aggregation is formation of number of things into cluster.

Carrier Aggregation in LTE:- 3GPP in Rel10 standardizes LTE-Advanced in addressing to meeting the IMT Advanced requirements. LTE-Advanced involves a set of features and one of them is carrier aggregation. In carrier aggregation it combines multiple LTE system bands or carriers with different bandwidth (1.5, 3 , 5 , 10 , 15 , 20 MHz) , thus increasing the overall capacity and the overall network throughput. Each aggregated carrier is referred to as a component carrier (CC) and the release 10 specifies that a maximum of five CC can be aggregated, hence a maximum aggregated bandwidth of 100 MHz, which is shown below.

 Carrier Aggregation scenario:- There are mainly 3 types of carrier aggregation depending on the component carriers.

1. Intra Band Contiguous(Continuos)
2. Intra Band Non Contiguous
3. Inter Band Non Contiguous

General Working Principal in Carrier Aggregation:- When carrier aggregation is used there are a number of serving cells, one for each component carrier. The coverage of the serving cells may differ, for example due to that CCs on different frequency bands will experience different pathloss. We have used only one carrier called primary carrier for coverage and all other carriers called secondary carriers are used for User data. The RRC connection is only handled by the Primary serving cell, Secondary serving cells,The SCCs are added and removed as required, while the PCC is only changed at handover.

Example: Suppose if we have 3 different Bands( 850,1800,2100MHZ) and we want to implement the CA(Carrier aggregation) then the principals as below:

       The primary cells/carriers must be 850 as we have less pathloss and more coverage and other 2 we can used in secondary cells/carriers for user data. 

Supervised Machine Learning - Overview and Examples

                                                        Supervised Machine Learning

Supervised Machine Learning is a type of machine learning in which we have both input and desired Output data are provided. Input and Output data are labelled for classification and regression to provide a learning  so that we can predict an output in future.

How does Supervised Machine Learning Work?- In Supervised learning we have an Input variable X and an Output variable Y and we use an algorithms to find the mapping function from the input(Input may be more than 1) to output as below:-
                                                        Y=f(X)
The Ultimate goal is to approximate the mapping function so that if we have new input data(X) we can predict an output(Y).

Why it is called Supervised?- Because the learning algorithm from the training data sets can be thought as a teacher supervising the class. The algorithms makes predictions from the training data sets and is corrected in case of wrong prediction. Learning stops when we achieved acceptable performance. 

Algorithms used in Supervised Machine Learning:- There are lots of Algorithms used. Some of are listed below:-
1. Linear Regression
2. Logistic Regression
3. Support Vector Machine
4. Decision tree
5. K-Nearest Neighbor
6. naive bayes.

Real world Examples of Supervised Machine Learning:-

1. To predict the price for House:- Suppose we have input data X(square footage, number of bedrooms, number of floors, Area, Year of build ) we can predict an output Y(price of that house).

2. To Predict the price of an specific stock.

3. To Predict the Loan Availability of an customer on basis of their credit history.

4. Spam email filter.

5. To predict an increase in Cell Traffic(In case of Telecom).

There are so many examples in real world. Wherever we have labelled input  and we have to predict an output we can use Supervised Machine Learning.