Sie sind auf Seite 1von 32

Optimizing UMTS Network Rollouts to Achieve CAPEX Savings While Enabling Long-term Network Performance

Mobile Network Optimization Africa - March 5, 2008 Tolga Kurt, Product Manager
1

Agenda
Network dimensioning Business considerations Technical considerations Network planning Coverage-focused approach Capacity-focused approach Mixed approach Network rollout 1:1 UMTS to GSM overlay Progressive overlay Coverage/capacity tradeoff Conclusion
2

Part 1: Network Dimensioning


Finding the site count that delivers optimal profitability

Business Objectives

Network Cost

Network Performance

Network Dimensioning
Business Considerations (1/2)
Many factors contribute to the required site count for a 3G network
Link budget Coverage target Optimization latitude
Profitab bility

An optimal operating point for a network does exist


It is highly influenced by the offered services and the network load As the network grows, more sites are required to achieve optimal profitability

Dimensioning is about finding the initial site count that will lead to maximal profitability while preserving long-term growth potential

Network Dimensioning
Business Considerations (2/2)

The key factors to assess during the business planning and dimensioning exercise are:
The coverage objective
The services to be provided The minimal coverage probability for each service The market penetration for each service The targeted service areas

The technology used to deliver high data rate services (e.g. HSDPA or not) The optimization latitude (i.e., separate UMTS antennas or not)

It is important to analyze the impact of these factors on the site count to establish the cost & value of these variables
All these factors influence the network dimensioning
5

What is a Coverage Objective?


100.00% 90.00%

Service Penetration

80.00%

70.00%

60.00% V oice 64kbps 50.00% 64/128 kbps 64/384 kbps 40.00% 0 5 10 Site Count 15 20 25

Implicit coverage objective of a 1:1 re-use

Impact of Design Decisions on Site Count


500 450 400 350 300 250 200 150 100 50 0 No optim ization E lectical tilts Service Penetration 95% S ervice P enetration 90% S ervice P enetration 85% M echanical tilts Az im uth

Coverage and Capacity Considerations


Given that the link budgets are reasonably compatible between GSM1800 and UMTS, the average cell spacing should be adequate It is extremely optimistic to think that a simple 1:1 re-use will deliver a network with good performance
Almost certainly, capacity cells do exist in urban areas Excessive propagation can be solved in GSM with proper frequency planning; this is not as easy in UMTS In fact, hierarchical cell structures are often used to provide a trade-off between coverage and capacity; such structures do not work in UMTS When enough spectrum is available in GSM, there is less emphasis on cell optimization and dominance of the main server

Key initial problem with a 1:1 re-use is the presence of a small (but important) number of substantial polluters, leading to lower capacity and quality

Excessive Pollution Problems

Ec/Io Plot efficiency Plot Spectral

Large areas with very low spectral efficiency and poor Ec/Io

1:1 Overlay Approach

Coverage Considerations of a Loaded Network


UMTS link budgets are very dependent on the actual network loading (i.e., cell breathing effect) During busy hours, the high data rate coverage will therefore suffer quite substantially (link budgets of a loaded UMTS network offering high data rate service is substantially worse than GSM) UMTS Voice link budget typically remains better GSM even when loading is quite high This translates to a network with solid voice coverage and fluctuating high data rate coverage
Might make it difficult to sell the value of high-data rate services when they are only solid during the quiet times

10

Impact of Cell Breathing


10% Network Load 75% Network Load
Probability of Service

Voice

> 99% > 95% > 90% > 50%

144 kbps

1:1 Overlay Approach

11

Impact of Cell Breathing: Statistics


Service Coverage Comparison
100.0 90.0 80.0 Percentage of calls 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0 Voice 10% Voice 75% 144 10% 144 75% 384 10% 384 75% Service & Loading
12

below 50 50 to 75 75 to 90 90 to 95 95 to 99 99 to 100

Part 2: Network Planning


Creating a network that delivers on business objectives
Initial Network Cost

Demand Anticipation

Network Longevity

13

Planning for Coverage


Initial network launch is performed with a network that provides good coverage but limited capacity
Considering the cell-breathing effect, the link budget is 3-6 dB better under low-loading conditions Spectral efficiency is not a true concern when the network load is low

Advantages
Substantial reduction in the site count (30%+) Lower initial CAPEX Faster rollout

Disadvantages
Higher cost in the medium-to-long term (there is a need for more sites) Network quality management is much more of a concern (there is a need for ongoing network optimization and the addition of sites) Network performance (high data rates) not as good at network launch

14

Planning for Nominal Loading Conditions


Inter-site distance (and site density) is heavily influenced by the link budget Nominal design must consider nominal loading conditions (e.g., a 60% load) for mature network This implies that sites are selected based on their ability to perform well together in a loaded network Advantages Investment associated to RAN is long-lasting (i.e. no need to decommission sites due to poor initial site selection) The network is capable of dealing with increasing traffic without material impact on the KPIs or need for further investment (i.e. additional sites) CAPEX utilization is optimal for normal loading conditions Disadvantages Initial CAPEX investment associated with network launch is materially higher (30%+) than if coverage is the only target Relies on good (and difficult) anticipation of long-term demand
15

Impact of Network Loading on the Site Count


100.00% 90.00% 50% Load Voice S ervice P enetration 80.00% 50% Load 64kbps 50% Load 64/128 kbps 70.00% 50% Load 64/384 kbps 25% Load Voice 25% Load 64kbps 60.00% 25% Load 64/128 kbps 25% Load 64/384 kbps 50.00%

40.00% 0 5 10 Site Count 15 20 25

Site count difference 64kbps service, 50% vs. 25% loading For a penetration target of 80% indoor

16

Achieving Both Goals (Mixed Approach)


The concept is simple: plan for the long term, deploy when the time is right
Network planning is performed based on normal operating conditions during peak hours (50-75% load) Initial launch is then performed based on subset of the sites selected

Advantages
Low initial site count (almost as low as coverage-based planning) Sites are known to be good contributors for the final network (i.e., investment is protected) Incremental additions to network are known to improve the network and to lead to an optimal network Less dependence on accurate forecast of long-term demand

Disadvantages
Increases the rollout complexity Requires software-aided processes to be achievable on a large scale

17

Network Site Count vs. Network Loading

Site Coun nt

18

Part 3: Network Roll-Out


Rolling-out a Network Efficiently

Optimize

Negotiate

Integrate

19

3G Rollout Philosophy: 1:1 Overlay


For each GSM macro-cell, a UMTS cell is deployed Advantages
Easy to rollout (almost no rollout considerations) Easy to outsource (simple design criteria) Works reasonably well if there are no GSM capacity cells

Disadvantages
Higher initial CAPEX Lower performance levels Does not work with all sites some GSM sites just cant be re-used More complex and expensive to optimize after the launch Makes it difficult to have various coverage objectives (no control over it)

1:1 Overlay Approach

20

1:1 Rollout Process


Install all Node B of a cluster For each GSM BTS for a given cluster, a UMTS Node B is installed. If dual-band antennas are installed, then the existing antennas are replaced with new dualband antennas. When a group of Node B (cluster) is enabled, various system parameters (GSM and UMTS) must be updated. This is particularly true for neighbor relationships in order to provide good handover performance. Most vendors rely heavily on drive tests as the impetus in the manual optimization process in order to achieve agreed KPIs. Drive tests can also be utilized to improve predictions when using an automated network optimization solution (ACP). Antenna configurations, power settings and neighbor relationships can be optimized based on drive-test data and/or predictions. Automatic or manual techniques can be used.

Integrate all new Node B

Coverage/Quality objectives are truly only dealt with during the optimization phase

Drive test area

Optimize radio parameters

1:1 Overlay Approach

21

Solving the Radio Pollution Problem


Reducing the amount of radio pollution is key to both coverage and capacity
Spectral efficiency is directly linked to the dominance of the best serving sector (i.e., the focus on in-cell noise rather than out-of-cell noise)

With a 1:1 re-use model, physical antenna parameters are the only efficient means of reducing radio pollution This translates into very significant post-design optimization efforts Given that propagation characteristics are only somewhat influenced by antenna parameters, it also means that the optimization process will not be able to improve performance as much as it could if the initial sites where selected on the basis of their radio characteristics Ultimately, there are cases where the only way to correct major pilot pollution problems is to shut down the polluter
This is something to avoid post rollout
1:1 Overlay Approach
22

Multi-band Optimization Considerations


Despite limitations, there are essentially three avenues
1. Optimize the UMTS and GSM1800/GSM1900 networks -- and possibly even GSM900 -- simultaneously (co-optimization)
Makes the optimization quite complex Makes the rollout of optimization changes difficult (need to coordinate between two technologies and multiple bands)

2. Optimize the UMTS network and assess GSM performance to manage coverage impact
This makes sense given there is a reasonable correlation between the performance of a UMTS network and a GSM network (i.e. making one work better leads to the other one working better) The primary factor to look for is down-tilting since UMTS optimization will typically lead to stronger downtilts than GSM would like (soft handover and need to reduce pilot pollution leads to more aggressive down-tilting in UMTS)

3. Optimize UMTS layer with constraints on the optimization


This is the most pragmatic approach provided that the operator is left with some level of optimization latitude for the UMTS network

1:1 Overlay Approach

23

3G Rollout Philosophy: Progressive Overlay


Site selection & density is directly driven by the coverage objectives
Density is therefore as variable as the coverage objectives Site selection and their parameters are based on their ability to contribute positively to the 3G network (i.e., doesnt assume a 2G site is a good 3G site)

Advantages
Lowest possible CAPEX/OPEX cost Worst polluters are inherently not re-used Allows for better control of the coverage objectives Provides a better method to deal with site availability issues Can deal better with a dynamic environment (i.e., problems with site acquisition, optimization constraints) Site negotiation is prioritized based on the true RF value of sites More complex process requires software-aided processes and possibly better skills

Disadvantages

Progressive Overlay Approach

24

Progressive Overlay Rollout Process


Define coverage objectives (Marketing) Service areas are defined by the Marketing Group with specific service objectives in mind (E.g. 90% 144 kbps indoor coverage for Metro Area)

Possible candidate sites would typically consist of the existing GSM sites + other available sites

Define a list of potential candidate sites within service area (140 sites)

Verify availability of the most valuable sites Define new list of most valuable additional sites (20 more sites)

Define the initial configuration likely to be used and the constraints for all possible candidate sites

Create nominal configuration for all candidate sites

Estimate CAPEX requirements (i.e., the site count) to fulfill the coverage objectives while considering design constraints

Determine number of required sites (100 sites) Have we reached coverage objectives?

Re-assess nominal plan and confirm sites required (which sites and how many)

Output: reference scenario Based on the reference scenario, determine the 20 sites that provide the best overall coverage contribution Establish a short list of the most valuable sites (20 sites)

No

Yes

Progressive Overlay

25

Site Potential Analysis


Site Name U_T04MX U_T04MCX U_T04MDX U_T04MKX U_T04MMX U_T04MVX U_T04MWX U_T23MJX U_T23MNX U_T34MQX U_T50MX U_T50MBX U_T50MQX U_T50MSX U_T73MLX U_T79MAX U_T80MX U_T82MX U_T86MAX U_T86MJX U_TB05MX U_TB14MX U_TMZZX Count 19 10 12 14 5 12 17 21 12 12 14 13 12 12 13 13 18 16 12 13 1 3 12 Value High Medium Medium Medium Low Medium High High Medium Medium Medium Medium Medium Medium Medium Medium High High Medium Medium Low Low Medium
26

Process Illustration
(Set of Possible Candidate Sites)
Possible candidate site (GSM site or other possibly available site)

Progressive Overlay

27

Process Illustration
(Creation of initial reference scenario)
Site selected in initial reference scenario (which achieves the coverage objectives)

Progressive Overlay

28

Process Illustration
(A site is not available, update reference scenario)

New site required due to a site not being available

Site is not required anymore

Site is not available anymore

Progressive Overlay

29

2-phase Rollout Process: Coverage/Capacity Tradeoff


Plan network based on nominal loading conditions Sites are selected on the basis of their ability to contribute to the coverage of a loaded network. Progressive overlay approach used for optimal results

Only sites that are part of the nominal design are considered for the initial rollout

Determine number of required sites for 20% loading

Plan network for initial launch based on a lower site count

Antenna parameters for the initial launch are most likely different than the ones in the final network

New sites are known well in advance and their selection is known to produce an optimal CAPEX/Performance ratio

Increment site count when network load increases

Coverage/Capacity Tradeoff

30

Conclusion
There are many variables to consider when planning the rollout of a UMTS network
Some of these variables directly impact the profitability of the network

There are ways to deploy a UMTS network to achieve the required network performance while minimizing the required expenditure Software solutions are enabling the efficient rollout of 3G networks, leading to better CAPEX efficiency for both the short and long term
31

Questions

32

Das könnte Ihnen auch gefallen