QoS and QoE Management in UMTS Cellular Networks Part 6

UMTS蜂窝网络QoS和QoE管理讲义

6.Radio Network

Dimensioning and Planning

Dr. David Soldani

(, tel. +358.50.3633527)

S-38.3215 Special Course on Networking Technology for Ph.D. students at TKK

UMTS蜂窝网络QoS和QoE管理讲义

Outline

WCDMA radio dimensioning and planning

Radio dimensioning aspects for UTRAN FDD A virtual time simulator for UTRAN FDD HSDPA dimensioning

(E)GPRS dimensioning

Procedure for CS and PS traffic

Dimensioning with capacity and bite guarantees Dimensioning with QoS guarantees

UMTS蜂窝网络QoS和QoE管理讲义

Dimensioning aspects for UTRAN FDD

3GPP leaves the engineering process, as planning and dimensioning aspects, to convert traffic demand into needed resources to vendors’and operators’choice

Hence, due to the complexity of the system and related expenditures, any

practical realization and deployment of new application servicesneeds to be validated a prioriby means of analytical approaches, or simulations, depending on the desired level of accuracy

None of the published analytical methods and tools showed enoughflexibilityfor an efficient and effective WCDMA radio interface dimensioning

We propose plain methods for radio interface dimensioning an a simple tool, which supports models for packet and circuit switched services, and processes snapshots of the system status, upon which performance statistics are derivedThe proposed solution is used to analyze the deployment of PoC (Push to talk over Cellular)by means of several case studies

UMTS蜂窝网络QoS和QoE管理讲义

Simulator structureStart

TG

Compute probability of a user to be active, generate active users even based and map bearers into services Scan inactive users Increase number of users, the probability of a user to be active is held constant Sort queue based on priorities Compute total loading for GB services and do the admission control Scan queue

AC

Outer loop: To Outer loop: To derive the derive the maximum load maximum load the cell can the cell can offer at given offer at aagiven QoS QoSN% of satisfied users< thresholds? Y Print offered and served traffic, throughput and% of satisfied users4

Compute load budget for NGB services Allocate bit rates to NGB users, i.e. do packet scheduling Scan queue Collect statistics

PS

Inner loop: The Inner loop: The new configuration new configuration of active users is of active users is repeated to repeated to approximate the approximate the desired result desired result more closely more closely

N All iteration done? Y

Compute% of satisfied users

PM

S-38.3215 Special Course on Networking Technology/ David Soldani/ Fall 2006

UMTS蜂窝网络QoS和QoE管理讲义

Call generator

Supported traffic models

CS speech and video calls

PS SWIS, PoC, streaming, MMS, WAP and dialup connections

Service (i) based parameters

QoS Profile: TC, THP, bit rates, ARP Mean service time (s): Ti Mean arrival rate (s): Ai

Share of subscription (%): Si

Probability Pifor service ito be used is:Group factor for PoC in Picomputation

Geometric distribution

Min =1, Max = 25, Mean =4

TiSi

Pi=

Ai100

Inner loop: Conditional probability for a user to make other callsOffered traffic in number of subscriptions Niis estimated as:Ni

Ui= average number of active bearers carrying the service i

Ai

=Ui

Ti

UMTS蜂窝网络QoS和QoE管理讲义

RRM and load estimates

GB is blocked if either one of the following in-equation is satisfied:

LTotal=LNGB+LGB>LTarget+Offset

LGB+ΔLGB>LTarget LNBGBcapacity

NGB traffic is always admitted, and bit rate allocated based on

PBNGB=LTarget (LNGB+LGB)

The load estimates are based on the fractional load equations

η

k=1+SHODL

W

ρkRkvk((1 α)+iDl)where

L∑k

Total=kηDL=PTxTotalPTxMax

UMTS蜂窝网络QoS和QoE管理讲义

Performance monitoring

User satisfaction criteria

not blocked and the average bit rate during the iterations (inner loop) is ≥to 8 (or 16), 64, 32, 64 and 32 kb/s, respectively

At a given traffic mix, the offered load is increased till at least one of the following conditions results true:

Method 1 (maximum offered load):

Less than 70%of MMS or WAP users are satisfiedLess than 50%of dialup users are satisfied

Less than 90%of users of any of other services are satisfied

Method 2 (impact of a new service on the existing subscribers satisfaction):

The subscription level is increased gradually for the new service while keeping the input load of the other services constant

UMTS蜂窝网络QoS和QoE管理讲义

Simulation Assumptions (1/2)

UMTS蜂窝网络QoS和QoE管理讲义

Simulation Assumptions (2/2)

* The volume is increased from 0 to 100%, whereas the average PoCgroup size is held constant: 1 user in Case 1 and

Case 2, 4 users in Case 3 and Case 4.

UMTS蜂窝网络QoS和QoE管理讲义

Case studies on PoC deployment

Case 1

Maximum allocated bit rate = 8 kb/s

Average PoC group size = 1 (one to one communication) Only different priorities are allocated to distinct services Method 1

Case 2

Maximum allocated bit rate = 16 kb/s All other settings as in Case 1 Method 1

Case 3

PoC av. group size in the same cell = 4 All other settings as in Case 2 Method 1

Case 4

Method 2

500 non-PoC users (held constant) All other settings as in Case 3

UMTS蜂窝网络QoS和QoE管理讲义

Case 1: Simulation results (1/2)

Served PoC traffic as a function of PoC subscriptions

60Cell throughput due to PoC users (kb/s)

5040302010

1

0%

8%

20%

40%60%

80%100%

PoC subscribers

0%

8%

20%

40%

60%

80%100%

PoC subscribers

Average PoC cell throughput < 50 kb/s≤7% of the total traffic in the cell

87PoC traffic in the cell (%)

65432

UMTS蜂窝网络QoS和QoE管理讲义

Case 1: Simulation results (2/2)

Impact of PoC traffic on other services

Insignificant: Only 7% of the other services would be not satisfactory if all the end users subscribed to PoC

Variation of max. n. of non-PoC subscriptions (%)

10099989796959493929190

0%

8%

20%

40%Max. number of subscriptions

0%

8%

20%40%60%PoC subscribers

80%

100%

60%80%100%

UMTS蜂窝网络QoS和QoE管理讲义

Case 2: Simulation results (1/2)

Served PoC traffic as a function of PoC subscriptions

10090Cell throughput due to PoC users (kb/s)

80706050403020100

0%

8%

20%

40%

60%

80%

100%

PoC subscribers

Average PoC cell throughput < 90 kb/s≤13% of the total traffic in the cell

1412PoC traffic in the cell (%)

1086420

0%

8%

20%

40%

60%

80%

100%

PoC subscribers

UMTS蜂窝网络QoS和QoE管理讲义

Case 2: Simulation results (2/2)

Impact of PoC traffic on other services

More significant than in Case 1: about 13% of the other serviceswould be not satisfactory if all the end users subscribed to PoC

Variation of max. n. of non-PoC subscriptions (%)

100999897969594939291908988878685

0%

8%

20%

40%PoC subscribers

60%

80%

100%

1100Max. number of subscriptions

10009008007006005004003002001000

0%

8%

20%

40%

60%

80%

100%

PoC subscribers

UMTS蜂窝网络QoS和QoE管理讲义

Case 3: Simulation results (1/2)

Served PoC traffic as a function of PoC subscriptions

Average PoC cell throughput < 250 kb/s≤1/3 of the total traffic in the cell

3530

250

Cell throughput due to PoC users (kb/s)

200

PoC traffic in the cell (%)

252015105

150

100

50

0%

8%

20%

40%

60%

80%

100%

PoC subscribers

0%

8%

20%

40%

60%

80%

100%

PoC subscribers

UMTS蜂窝网络QoS和QoE管理讲义

Case 3: Simulation results (2/2)

Impact of PoC traffic on other services

Worst case: Significant deterioration of the performance of other services if more than 20% of the end users subscribed to PoC

Variation of max. n. of non-PoC subscriptions (%)

11001000Max. number of subscriptions

9008007006005004003002001000

0%

8%

20%40%60%

80%

100%

PoC subscribers

UMTS蜂窝网络QoS和QoE管理讲义

Case 4: Simulation results (1/2)

Average cell throughput as a function of PoC subscriptions

As expected, when the PoC traffic increases the NGB load decreases (PoC has higher priority), whereas the load due to GB services remains constant (PoC has no means to affect the AC of GB services)

700600Cell throughput (kb/s)

5004003002001000

0%

8%

20%

40%PoC subscribers60%

80%

100%

UMTS蜂窝网络QoS和QoE管理讲义

Case 4: Simulation results (2/2)

Impact of PoC traffic on other services

Significant deterioration of WAP performance if more than 50% ofthe end users subscribed to PoC: Extra capacity needed in the cell

1009080Satisfaction level (%)

706050403020100

0%

8%

20%

40%PoC subscribers

60%

80%

100%

UMTS蜂窝网络QoS和QoE管理讲义

Virtual time simulator for UTRAN FDD

In UMTS only a layered bearer service architecture and QoS attributes are defined: Implementation and planning aspects of the actual QoS management functions are left to vendors’and operators’choice

Due to the complexity of the system and infrastructure costs, any practical deployment of radio resources management (RRM) algorithms and offered services in UTRAN needs to be validated a prioriby means of static or dynamic simulations, depending on the desired level of time resolution and accuracyWe present a virtual time simulator that overcomes the

limitations(snap shot of the system status only) and complexity(far too high time resolution) of static and dynamic system level simulators

UMTS蜂窝网络QoS和QoE管理讲义

Simulator structure

Modular structure with clear interfaces: Each module is implemented independently so that each entity may be straightforwardly replaced by an alternative solution

Supported functions: Traffic and path loss generators, Admission

Control (AC), Load Control (LC), Packet Scheduler (PS), Power Control (PC), Process Calls (PrC) and Performance Monitoring (PM)

Mobility effects and SHO gains:may be taken into account by e.g. speed dependent Eb/N0 requirements and SHO condition

DHO branches are processed first followed by the main branches, the bit rate assigned to the radio link set (UE) is the minimum of the bit rates allocated separately (for each cell) to all radio links of the active set

The maximum resolution of the tool is one radio resource indication period (RRI), i.e. the time needed to receive the power levels from thebase stations

UMTS蜂窝网络QoS和QoE管理讲义

Simulation flow chart

本文来源：https://www.bwwdw.com/article/enm1.html