Guideline for Network Design and Optimization

Guideline for Network Design and Optimization

4.2.2.5

Access Failure Rate (1-Call Setup Success Rate):

The proportion of call setup attempts that fail.

Access Failure Rate % 12.0% 9.8%

% 10.0% of Ac ce 8.0% ss Att 6.0% em pts 4.0%

8.2%

2.0% 0.0% Network A

4.2.2.6

Network B

Blocked Call Rate:

The proportion of call attempts that fail due to lack of resources. Blocked Calls and No Service [%] 12% 9.5%

10%

8.0% 8% 6% 4% 2% 0.3%

0.2%

0% Network A

Network B No Service Attempts

Blocked Calls

Page 42


4.2.2.7

Call Drop Rate:

The proportion of calls terminated abnormally before the end of the call.

Dropped Call Rate % 3.5% % 3.0% of Co 2.5% mp let 2.0% ed Cal 1.5% ls

2.9%

1.3%

1.0% 0.5% 0.0% Network A

4.2.2.8

Network B

Handover Failure Rate:

The proportion of handover attempts that fail.

Handover Summary 1600

12

1400 Nu m be r of Ha nd ov er

1200

35

1000 800 600

1444 1176

400 200 0 Network A

Network B Handover c omplet ed

Handover failed

Page 43


4.2.2.9

Average SQI:

The average Speech Quality Index measured over the combined drive test route.

Average Speech Quality Index (SQI) 24 21 SQI

19.3

19.1

18 15 12 9 6 3 0.0 Network A

Network B

4.3 GPRS Drive Test GPRS drive test data can be presented in a number of ways, much the same as GSM drive test data. A combination of graphical presentation and statistical analysis is recommended.

4.3.1

Graphical Presentation

The following parameters can be displayed on a map, allowing the visualisation of specific problems by location: 4.3.1.1

Route Plots

UL/DL RLC Throughput:

Radio Link Layer data throughput

UL/DL LLC Throughput:

Logical Link Layer throughput (user data)

UL/DL RLC Block Error Rate (BLER):

Radio Link Block Error Rate

UL/DL RLC Retransmission Rate:

Radio Link Retransmission Rate

UL/DL Coding scheme used (CS1-4):

Allocated Coding Scheme

UL/DL Number of timeslots used:

Allocated timeslots

Page 44


4.3.1.2

Events

Events plots may be superimposed on one of the available route plots, eg. RxLev, RxQual, RLC throughput, etc.

PDP Context Activation Failure:

Failure to activate PDP Context (Packet Data Protocol)

PDP Context Loss:

Loss of PDP Context (GPRS Call Drop)

4.4 Network Performance Review - Summary The summary of the Network Performance Review should aim to highlight the specific performance problems identified in the network, on Network level, BSC level and Cell level. The following headings should be included here: •

Network Performance Summary Data

Network Name

XYZ-net

e t a R s s e c c u S l l a C

91.70%

e t a R s s e c c u S p u t e S l l a C

93.40%

e t a R l l a C p o r D

1.85%

•

Key Network Performance Observations

•

List of worst performing cells and BSC’s

n o i t s e g n o C H C T

0.73%

t a R s s e c c u S t n e m n g i s s A H C C D S

92.10%

c i f f a r t / e m u l o V l l a C

1244300

e t a R s s e c c u S r e v o d n a H

95.60%

Detailed conclusions can be made only after completing the Network Design and Dimensioning Review, at which time all the required information will be available to allow detailed recommendations to be made.

Page 45


5

N ETWORK D E S I G N A N D D I M E N S I O N I N G R E V I E W

5.1 Network Design Summary Before making any recommendations based on network performance reports it is important to know more about the network, and the constraints inside which the network has been designed and is being operated.

5.1.1

Size

How big is the network? Plots from network planning tools are useful as a visual aid, along with numerical information in spreadsheets: •

MSC’s

•

BSC’s

•

BTS’s

•

Cells

•

OMC’s

•

HLR/VLR,

•

SMS Centres

5.1.2

Subscribers

Subscriber Distribution, usage and growth information: •

Roughly how many subscribers distributed over the network, by area or by clutter.

•

Projected subscriber growth, pre-paid and fixed contract.

•

Traffic generated by subscriber, current and projected (typically in the range of 2025mE per subscriber in the busy hour)

5.1.3

Description of the environment

It is helpful to know about environmental factors that influence network design and performance, such as: Type of urban environment (typical building heights, building density, etc.) Type of terrain (mountainous, hilly, flat, etc.) Presence of water bodies (coastline, estuaries, rivers, lakes)

5.1.4

Available Spectrum

What spectrum is available, and how is it split between the different layers?

Page 46


The following example shows a typical allocation of GSM channels given an available spectrum of 10MHz (50 Channels).

Guard Band

Guard Band

1 ch

1 ch

BCCH

TCH Hopping

MICRO

14 ch

26 ch

8 ch

Dual Band (900/1800) spectrum should also be shown.

5.2 RF Design Detailed Analysis The high level design summary provides an overview of the relevant information. Next a more detailed analysis is required.

5.2.1 5.2.1.1

Site Design Network Growth Pattern

Networks in urban areas (especially older networks) tend to follow a set growth pattern: •

Launch rollout with minimum sites for maximum coverage.

•

Fill in coverage holes and add capacity by cell splitting

•

Add increasing numbers of microcells, in-building cells and street-level cells to increase capacity focused on high subscriber density areas.

In terms of RF design, the problem with this approach is that the legacy sites from the launch rollout phase tend to be high and prominent, and increasingly contribute uplink and downlink interference into the network as the number of lower sites around them increases. The net effect of this is to minimise frequency re-use efficiency and limit the capacity of the network. Therefore a process is required to identify and eliminate these interferers to allow network growth to continue and high quality to be maintained. 5.2.1.2

High Sites Replacement

A typical process for replacing or modifying high sites would be as follows: •

From BSS performance statistics and call trace logs, identify those cells which contribute the most interference to the largest number of other cells.

Page 47


•

Develop a plan for de-commissioning the site, or lowering the antennas to a position consistent with surrounding sites if possible. Include in the rollout plan the requirements for additional in-fill sites due to the loss of coverage from the high site.

•

As new low sites are integrated, de-commission or modify the high site in such a way as to cause minimum disruption to coverage. Prioritise the integration of the required new sites to target high sites in order of severity.

The network design review will include a study of high sites in urban areas of the network where growth is limited by frequency re-use problems. An action plan will be developed according to this outline process, and will be provided as an input into the network expansion and rollout process. 5.2.1.3

RF Design Strategy

Although not strictly part of a performance and optimisation review, it is important to consider the design strategy in place in the network, and to provide input into the expansion process to account for performance-related issues. This includes a review of the following design techniques: •

Microcellular and Picocellular underlay

•

Dual Band (Dual-BCCH and Single-BCCH)

•

In-building cell deployment

5.2.2

Traffic Distribution

In most networks it is found that the distribution of traffic between cells is not even, and that a small number of cells may be heavily congested while most others are underutilised. The key to the efficient utilisation of network infrastructure is to attempt to distribute traffic evenly between BTS’s and achieve maximum frequency re-use efficiency. There are various techniques available to achieve this, including: •

Removal of high or prominent sites which tend to ‘suck in’ disproportionate levels of traffic owing to their high coverage level compared to surrounding sites.

•

Downtilting antennas to reduce levels of unwanted coverage outside the intended coverage area.

•

Hotspot detection: Using Call Trace logs, it is possible to determine roughly the location of traffic hotspots, helping the RF designer to plan new sites in exactly the right locations to serve high traffic areas. This also has the effect of reducing the average path loss between BTS and mobile (because on average the BTS’s are closer to the mobiles), and therefore the interference levels in the network are reduced.

•

Traffic Management Algorithms: Many BSS vendors provide advanced traffic management algorithms, allowing traffic distribution to be controlled to a greater extent by the optimiser. These included microcell handover algorithms, congestionbased handover algorithms and so on.

Page 48


The design review will include a study of the traffic distribution across the network, and for the most congested cells recommendations will be made for ways to re-distribute traffic. In many cases these inputs are directly relevant to the ongoing network expansion and rollout process. This will also include a review of GPRS traffic projections, and how this will impact the combined traffic distribution carried by the network.

5.2.3

Frequency Plan

Frequency Planning is a complex subject. The quality of a frequency plan (re-use efficiency, interference levels) is directly related to the quality of the RF design. A poor frequency plan is usually the result of a poor RF design, resulting in turn in an inability to produce a good frequency plan. This section attempts to highlight the main considerations behind creating an efficient frequency plan. 5.2.3.1

Site design

As mentioned in previous sections, frequency reuse efficiency is affected by site design. Inconsistent site heights (mixture of high and low sites) reduce re-use efficiency. 5.2.3.2

Terrain and Topography

Hilly terrain presents more frequency planning problems compared to flat terrain, as cell coverage areas are harder to control and unwanted ‘splashes’ of coverage are hard to avoid. Site design and antenna location can be critical in minimising these effects. 5.2.3.3

External Interference

Sometimes the performance of radio channels is affected by external interference (ie. interference originating from outside of the network). This could be due to unauthorised users occupying radio spectrum for other communications purposes. An example of this is the 900MHz cordless telephone standard used in the USA, that use part of the GSM Uplink spectrum (between channels 70 and 75). This is allowed in the USA but causes problems to mobile networks in other countries where these channels are licensed and allocated to GSM operation. Although these phones are generally not licensed to be used outside the USA, they are widely available in most countries of the world and result in strong uplink interference. Another example could be interference in coastal or port areas from radio communications systems offshore (such as shipping, drilling platforms, etc.). Finally, in border regions of neighbouring countries there may be spectrum re-use issues. These can generally be resolved by agreements between operators in the neighbouring networks. 5.2.3.4

BCCH Plan

The number of channels required to make a good BCCH plan will vary according to a number of factors: •

Site Design (high sites etc.)

•

Terrain and topography

•

Subscriber distribution Page 49


•

Regularity of cell plan

In a well optimised network, it is generally possible to produce a high quality BCCH plan within 14-15 channels. 5.2.3.5

Non-BCCH Plan

The same issues with the BCCH plan also affect frequency planning of the non-BCCH (TCH) carriers. However there are additional techniques available for the TCH layer to improve re-use efficiency and increase capacity, such as: •

Synthesizer Frequency Hopping

•

Baseband Frequency Hopping

•

MRP (Multiple Reuse Pattern)

•

Concentric Cell

These are described in detail in the ‘Optimising for Growth’ section. The network design review will include a study of the frequency plan, and will suggest optimisation steps required in order to produce a more efficient plan and hence a better quality and higher capacity network.

Page 50


5.3 Optimising for Growth The need for optimisation generally arises out of a need for growth and expansion of a network to serve a growing number of subscribers, and to support an increasing range of services. This section attempts to describe the optimisation techniques available for maximising network capacity while maintaining high network quality. The availability and effectiveness of these features and optimisation techniques varies between infrastructure suppliers. The network optimisation process can be represented in a diagram, as shown below:

Drive Test Data

A-Interface Data

Call Trace Data

OMC Stats Data

Performance Reporting

QOS Metrics

OMC Management

Field Operations

Database Parameters Quality-Driven Network Design Review, Expansion and Optimisation Process

Network Operations - Rollout - Change Control

RF Design Parameters

Expansion Plans

Core Network Design Parameters Marketing Strategy

Optimisation Plans

Performance Requirements

Network Planning and Optimisation

The effectiveness of all of these features also largely depends on the network design, and how the feature parameters are optimised. A careful examination of all design factors affecting the use of these features should be undertaken, and recommendations made as to the suitability of the features and/or improvements in performance through optimisation.

Page 51


5.3.1

Synthesizer Frequency Hopping (SFH)

SFH is a widely accepted technique in GSM for providing capacity and quality improvements. These benefits are as a consequence of the following features of SFH: •

Increased immunity to fading due to frequency diversity.

•

Better frame erasure rate through interference averaging

•

Greatly simplified frequency planning allowing faster rollout and better quality.

The effectiveness of SFH in achieving capacity and/or quality gains is dependent on a number of optimisation-related factors: 5.3.1.1

Hopping spectrum allocation

Since the benefits of SFH arise as a consequence of the nature of spread-spectrum operation, the amount of benefit is related to the degree of spreading. In SFH this is determined by the spread of channels allocated in the MA list (hopping sequence). Simulations show that up to around 2MHz spread (10 channels) there is an appreciable increase in hopping gain, but above 2MHz spread the additional gain reduces. 5.3.1.2

Choice of SFH Design

SFH can be deployed in a number of ways according to the network design. For example: 1x3 SFH:

In this scheme, the hopping band is divided into 3 equal groups and planned according to a regular re-use pattern. This is suited to networks with regular cell plan and 3-sector sites

1x1 SFH:

In this scheme, the whole hopping band is allocated to a single hopping group, which is re-used in every cell and every site. This technique is better suited to irregular networks.

1x1 Split SFH: This is similar to the 1x1 SFH scheme, except that it allows for operation with different cell layers (for example high sites and low sites). The hopping band is divided into two groups, and each group is applied according to the 1x1 scheme on a per-layer basis. Other variations are also possible, depending on the particular implementation of the technique in the supplier’s BSS software. 5.3.1.3

Hopping System Parameters

A full review of the use of hopping system parameters is required, to ensure compliance with recommended SFH planning rules. MA List:

Frequencies allocated to the hopping sequence

HSN:

Hopping Sequence Number (0 = cyclic, 1-63 = pseudo-random)

MAIO:

Mobile Allocation Index Offset. Sometimes set automatically, however manual definition of MAIO is essential for the correct implementation of certain hopping techniques (eg. 1x1 SFH).

These parameters also apply to a baseband hopping system, although their use is somewhat different.

Page 52


5.3.2

Baseband Frequency Hopping and Multiple Re-use Patterns (MRP)

This is a technique preferred by a few suppliers, notably Ericsson, although SFH is a more commonly used technique. MRP is a variation of baseband hopping in which frequencies are allocated to carriers hierarchically with an increasingly aggressive re-use pattern. In other words, the BCCH carrier would be planned with a 5x3 pattern, TCH1 with 4x3, TCH2 with 3x3, and so on. TCH channels are then allocated in priority order, starting with the BCCH. One feature of MRP is that since interference increases on the ‘higher’ carriers due to the increasingly aggressive re-use patterns, the area in which an acceptable C/I can be achieved those carriers is correspondingly smaller. This requires careful optimisation to maximise traffic capacity.

5.3.3

Downlink Power Control and DTX

Downlink power control is important in frequency hopping systems as a means of reducing downlink interference. Downlink PC parameters should be reviewed and, recommendations made if necessary. DTX (Discontinuous Transmission) is also sometimes used. This feature allows the BTS to use voice activity detection, and then to transmit idle frames during breaks in speech, thus reducing average downlink power and interference.

5.3.4

Microcell Traffic Management Algorithms

Some BSS suppliers provide advanced traffic management alkgorithms designed to control the distribution of traffic between different cell layers of the network. Typically this is applied between macrocell and microcell layers. Since microcell laters are usually designed to carry high capacity in small coverage areas, the principles behind these algorithms are generally as follows: •

Prevent handover from micro layer to macro layer unless the handover cause is imperative (Qual or Lev).

•

Prevent handover from macro layer to micro layer for fast-moving mobiles, to prevent unnecessary handovers and potential call drops due to handover failure.

•

Encourage handover from micro-to-micro and avoid handing back into the macro layer.

•

Encourage mobiles to remain camped onto to micro cells despite lower signal level through use of modified cell selection algorithm (C2).

The use of these algorithms must be reviewed to ensure optimum traffic distribution and correct handover operation.

Page 53


5.3.5

Dual Band Traffic Management Algorithms

In the case of dual band system operation, most suppliers provide traffic management algorithms to prioritise traffic channel allocation between 900 and 1800 layers. These include: •

Prioritisation of 900 or 1800 layer

•

Rules for assigning TCH in 900 or 1800 layers according to traffic loading

•

Single-BCCH operation allowing 1800 TCH allocation from 900 BCCH.

A full review of the use and effectiveness of these features is required, and recommendations made if necessary.

5.4 The Network Growth Planning Process Network growth planning requires a number of inputs. Network growth and expansion planning is covered in greater detail in separate documents, however the inputs into the process can be represented in a diagram, as shown below:

Marketing Strategy RF Design Constraints

Performance Reports

Capacity Requirements

Site Acquisition Constraints Network Design Review, Expansion and Optimisation Process

Coverage Requirements

Available Spectrum

Quality Requirements

Network Planning and Optimisation

Capacity and Performance enhancing features

Network Operations - Rollout - Change Control

A typical network expansion process will include all of these inputs as a minimum.

Page 54


5.5 BSS Database Review Many BSS database parameters are specific to equipment vendors, while some are defined by ETSI in the GSM specification. Even those which are ETSI-defined may be optimised slightly differently according to the vendor-specific implementation of software algorithms, so it is not possible to derive a single set of default values valid for all equipment vendors. The results obtained from the network performance review will tend to suggest which aspects of the BSS database may be able to be optimised. This section attempts to suggest which specific areas within the BSS database should be reviewed.

5.5.1

Radio Resource Timers

Many radio resource timers exist in GSM. Generally speaking, their main purpose is to ensure de-allocation of radio resources after the failure of some resource allocation process, and thus ensure maximum utilisation and minimum wastage of resources. The following timers are by no means an exhaustive list, but are commonly optimised to maximise resource utilisation. 5.5.1.1

rr_t3111 (layer 2 channel release guard timer) =>1200ms

This timer is used during the normal layer 2 channel release procedure. Its purpose is to maintain the dedicated channel in an active state long enough for the MS to repeat the L2 DISC message if required. rr_t3111 commences when the first DISC is received by the BSS, and up to 5 repetitions of the DISC are allowed, at intervals of T200 (235ms on SDCCH, 166ms on FACCH). This means that the maximum time that the dedicated channel needs to be held is 1175ms (5 x 235ms). A higher setting of rr_t3111 will hold SDCCH or TCH resources longer than necessary, possibly introducing SDCCH or TCH congestion. A lower setting could result in two mobile stations being active on the same channel. To avoid the possibility of 2 mobile stations active on the same channel and to safely minimise channel usage, it is recommended that rr_t3111 be set to 1200ms across all cells.

5.5.1.2

rr_t3212 (Periodic Location Update Timer) => Align With MSC Implicit Detach Timer

rr_t3212 is transmitted by the BSS in the BCCH System Information and is used by the mobile as the periodic location update timer. The mobile restarts the timer each time it successfully location updates. If the timer ever expires then the mobile makes a periodic location update. If the IMSI detach feature is enabled this timer should be set to a value less than the MSC implicit detach timer. If rr_t3212 is set higher than the MSC implicit detach timer then mobiles which are camped on the network will not be paged if they do not location update before expiry of the MSC implicit detach timer.

Page 55


It is recommended to set this timer to a lower value than the MSC implicit detach timer. The effect of this timer change will be to improve the mobile terminated call set up success rate measured by the MSC and perceived by anybody trying to call a mobile station from either the fixed network or another mobile station.

5.5.1.3

link_fail => 16 SACCH

The link_fail timer governs the number of missing uplink SACCH messages that should occur before a radio-link-loss is determined by the BSS. As the loss of uplink SACCH indicates that uplink audio is also lost, it is likely that the subscriber will have terminated the call after 16 SACCH multiframes have passed. This represents a time period of 7.7 seconds. The benefit of minimising this timer value is that the holding time of the channel following any radio link loss is minimised, while still allowing sufficient time for the link to recover before the subscriber gives up and terminates the call.

5.5.1.4

radio_link_timeout => 16 SACCH

Radio_link_timeout governs the number of missing downlink SACCH messages which should occur before a radio-link-loss is determined by the mobile. It is recommended to set this timeout to 7.7 seconds (16 SACCH) for all cells, in line with the link_fail recommendation above. It is a very small percentage of calls which would recover from a losing SACCH for longer than 7.7 seconds and the user is likely to have released the call anyway due to loss of audio. This will help to maximise the use of TCH resources by returning them to the radio resource pool as soon as possible, but without losing calls which might otherwise recover. 5.5.1.5

rr_t3109 (TCH Reallocation Timer) => 8000ms

This timer prevents the reallocation of a channel, following the detection of an uplink radio link loss. The MS may still be transmitting on the channel until radio_link_timeout expires. rr_t3109 should therefore be set to a value greater than radio_link_timeout. Based on a radio_link_timeout of 16 SACCH as recommended above, then it is recommended that rr_t3109 should be reduced to 8000ms for all cells. This will ensure that channels will be released as soon as is safely possible following a radio link loss, increasing channel availability, avoiding any unnecessary SDCCH or TCH congestion.

5.5.1.6

rr_t3103 (Intra-BSS Handover Guard Timer) => 15000ms

This BSC call processing timer is used during an intra BSS inter cell handover, to guard against allocation of resources to a handover after it has failed. To maximise the intra BSS handover success rate this timer should be long enough to allow the MS to receive the handover command over the air interface, try and fail to

Page 56


access the target channel and to return to the source channel, including LAPDm layer 2 repeats at each stage. The timer should not be so long that resources are held up unnecessarily introducing channel congestion. To maximise the intra BSS handover success rate it is recommended that this timer should be set to 15000ms.

5.5.1.7

bssmap_t10 (Assignment Guard Timer) => 14000

This timer runs at the BTS during the assignment procedure, guarding the non-receipt of the assignment complete or assignment failure message from the MS. On expiry of this timer, the radio channels will be released and a Clear Request message sent to the MSC This timer must be sufficiently long to maximise the probability of a successful assignment. In worst case conditions the MS will take about 13s to fail an assignment and recover to the source channel. Therefore the timer setting 14000 ms is recommended.

5.5.1.8

bssmap_t8 ( Handover Guard Timer) => 14000

This timer runs at the source BTS during an intra-BSC or inter-BSC handover. It starts when the Handover Command is transmitted, and is stopped when the BSC clears the source cell resources following a successful handover or when a Handover Failure message is received. If it expires, resources at the source BTS are released. Again, this timer needs to be long enough to allow the MS a reasonable chance to recover in poor radio conditions, but it also needs to be set slightly smaller than rr_t3103. It is recommended to set this timer less than rr_t3103 (15000 ms). Therefore if rr_t3103 is set to 15000 ms, as recommended, then bssmap_t8 should be set to 14000 ms.

5.5.2

Handover and Power Control Parameters

Handover and power control parameters are set according to the implementation of handover and power control algorithms on a vendor-specific basis. However there are some general guidelines and recommendation that can be applied independently of vendor-specific implementations. 5.5.2.1

RxQual Handovers:

It is recommended to configure settings such that a handover will be triggered when 4 consecutive measurement reports contain at least one RxQual value of 7, together with at least one RxQual value of 5 and two RxQual values of 6 (or some setting similar to this). RxQual handovers should take place relatively quickly to avoid potential loss of speech quality, but not too quickly such that excessive handovers take place along with a higher risk of call drop The RxQual handover should only take place if the BTS/MS are at full power and a target cell is available at an equivalent or stronger downlink RxLev as the server. If

Page 57


possible, different RxQual thresholds should be set for hopping and non hopping channels, since hopping channels can tolerate a worse BER for the same FER as compared to non-hopping channels. 5.5.2.2

RxLev Handovers:

RxLev handovers are generally not too useful. If a call is in progress and the quality is acceptable, there is no need to perform a handover purely based on a low RxLev. Similarly if the RxLev is low and the quality is poor, a handover can be performed based on RxQual. So there are very few situations in which an RxLev handover is useful, and the recommendation in most cases would be to keep them disabled. In case they are used, RxLev handovers should be considered only as a ‘last resort’, and the RxLev handover cause should be generated only when the RxLev reaches a very low value, and it should be checked that a handover to a weaker cell is not possible. 5.5.2.3

Uplink Power Control:

The basic philosophy of uplink power control is that the MS should be powered down while the received signal level at the BTS is of good quality (RxQual =0). If bad RxQual should occur (RXQual >0), the MS should power up again, until the RxQual improves. Generally speaking, power-down commands are given based on good level, while power-up commands are given based on poor quality. Power-up needs to be fast in order to quickly overcome quality problems by increasing power. Occasionally powerup commands will given based on low level. Typical power control thresholds are as follows: l_rxqual_ul_p =

56

(lower RxQual threshold)

u_rxqual_ul_p =

0

(upper RxQual threshold)

l_rxlev_ul_p

=

20

(lower RxLev threshold)

u_rxlev_ul_p

=

30

(upper RxLev threshold)

pow_inc_step_size

4

(power increase step size, dB)

pow_red_step_size

2

(power decrease step size, dB)

5.5.2.4

MS Fast Power Down:

Some equipment vendors provide a feature allowing the mobile to be quickly powered down immediately following call setup by a large step, provided the RxLev is above a defined threshold. A ‘feature’ of GSM is that mobiles transmit full power at call setup, and power down through normal power control thereafter. This process can be slow and results in mobiles transmitting higher than required power for a significant time. Fast power down algorithms should be used wherever available as this reduces average uplink interference in the network.

Page 58


Some vendors also offer a feature to set the mobile power at the required level on handover based on path loss measurements, rather than reverting the mobile to full power automatically, This should also be used wherever available.

5.5.2.5

Downlink Power Control:

Downlink power control is useful as a means of reducing levels of downlink interference, and should be particularly applied to cells known to be interferers (such as high or prominent cell sites possibly over-shooting lower cell sites). Parameter settings could be similar to those used for Uplink Power Control.

5.5.2.6

Adaptive Handover:

The principle of adaptive handover is to replace the usual fixed voting algorithm used for handover decision making with an algorithm based on ‘rate of change’ of a condition. In other words, if there is a sudden large reduction in RxQual the handover will be triggered more rapidly compared to a slow and gradual reduction. This represents an improvement over the standard voting mechanism (n out of p etc.), and is recommended to be used wherever available.

5.5.2.7

Adaptive Power Control:

Some equipment vendors provide an adaptive power control algorithm, which allows the step size of power control commands to be changed according to how far above or below the thresholds the measurements are. This enables a faster power control within the defined power thresholds and minimizes uplink and downlink interference.

5.5.2.8

Directed Retry and Intelligent Directed Retry (Handover on Congestion):

Most, if not all, vendors offer the Directed Retry feature, and some offer an enhanced version known by various names, but based on the principle of ‘Handover on Congestion’ The Handover on Congestion feature works as follows: •

Call setup attempt is blocked due to lack of TCH resources

•

Blocked call setup attempt is queued, occupying a SDCCH.

•

One of the existing calls in the congested cell is handed over to a neighbour cell, freeing a TCH.

•

The TCH thus freed is allocated to the call in the queue, and the call setup is completed.

The handover cause generated is ‘congestion’, and is handled in a similar way to a normal power budget handover. The handover has a margin associated with it, allowing the operator to control the conditions under which neighbours can qualify for Page 59


congestion handover. Handover on Congestion is generally more effective than normal Directed Retry for the following reasons: •

Directed Retry allows call setup to cells outside the intended coverage area, increasing the probability of poor quality.

•

Handover on congestion chooses the best candidate in the cell for handover based on a measurement history. Directed Retry has no equivalent measurement history.

Handover on Congestion is recommended as an effective method of limited redistribution of traffic in a congested network through improved trunking efficiency, and allows the network operator some enhancement in capacity during periods of rapid network growth.

5.6 Location Area Planning and Paging Performance Good location area planning should minimise the number of location areas in the network and minimise the amount of location update traffic. The main benefits of this are: •

Reduction in SDCCH resource usage through minimised location update traffic. The location of location area borders is also critical for this, as well as the number of location areas.

•

Improved Call Setup Success Rate: The mobile is more likely to be established on the best serving cell when away from a location area border, since cell_reselect_hysteresis applies only at location area borders (on location update) but not between cells of the same location area.

Page 60


The maximum size of a location area is limited by the available Paging Channel (PCH) capacity. The PCH capacity is fixed on a per cell basis by the CCCH configuration and access grant block reservation set in the BSS database by the parameters ccch_conf and bs_ag_blk_res respectively. The theoretical maximum paging capacity for each possible CCCH configuration is shown in the following table:

Paging capacity is the primary consideration for calculating location area size. Additional Location Area planning issues are as follows: •

Location Areas must be contained within an MSC, so MSC borders place a restriction on Location Area border planning

•

Geographical and topographical considerations impact location area planning, such as planning to minimise movement of mobiles across location area borders, avoiding borders following rivers through large cities (RF propagation is difficult to control over water, so results in many unplanned location updates), and so on.

A well-planned network should have similar paging loads in each location area with the maximum paging load within reasonable range of the theoretical maximum paging capacity. A very small paging load would suggest that the location area is too small and could be combined with neighbouring location areas, minimising location update activity and reducing use of SDCCH resources. A paging load too close to the Page 61


theoretical maximum paging load would suggest that the location area is too large and should be split up into multiple location areas to avoid paging overload. Location area planning should be reviewed by investigating the paging load per location area, and making recommendations as required: Paging load too low: Combine small location area with other adjacent location area(s) to reduce SDCCH usage and reduce unnecessary location update activity. Paging load too high: Consider splitting location area into smaller ones. Paging load OK but location update activity too high: Consider re-planning location update borders to minimise cross-border mobile movement and thus minimise location update activity. Look for ‘islands’ (cells accidentally hosted in the wrong location area causing unnecessary location updates) and coverage ‘splashes’ (especially near rivers, estuaries etc.).

5.7 System Processor Performance The loading of system processors should be checked in accordance with the planning and dimensioning guidelines provided by the equipment vendor. The following processors should be checked: •

BSC processors

•

BTS processors

Any overload conditions should be reported, and processor expansion/upgrade recommended as necessary.

5.8 MTL Performance The MTL C7 signalling links between BSC and MSC (A-Interface) are key system components to ensure high system performance, including call setup success. Therefore these links should be designed with high availability and dimensioned appropriately to avoid any overload or congestion conditions. Any observed overload conditions or outages should be reported, and action recommended as necessary.

Page 62


5.9 Additional BSS Design Issues The following additional BSS design issues should also be reviewed:

5.9.1

Hardware configurations

Networks often tend to consist of multiple generations of GSM hardware, as well as BSS supplied from several vendors. It is important to check hardware configurations to ensure they support the required quality and capacity features (such as synthesiser frequency hopping).

5.9.2

Transmit Combining Options

The choice of transmit combining method has a significant impact on coverage (due to insertion loss of the combiners), and the possibility to use certain capacity features (SFH, MRP, etc). Transmit combining should be considered as part of the BSS design review, and recommendations made if necessary.

5.9.3

Antenna Selection

Antenna specifications have a significant impact on network performance. The suitability of antennas should be reviewed according to the observed performance problems in the network, and recommendations made as necessary, eg: •

Vertical and Horizontal Beamwidth

•

Gain

•

Front-to-Back Ratio

•

Null Fill

•

Downtilt (electrical/mechanical)

Antenna positioning is also improtant with resopect to minimising interference and unwanted radiation. This should also be studied in relation to the RF design strategy.

5.9.4

Diversity Choice

The use of diversity should be reviewed. Lack of diversity can result in link balance problems, in turn resulting in poor call setup performance and poor quality. Different types of diversity are possible: •

Horizontal space diversity

•

Vertical space diversity

•

Polarisation diversity

Sometimes diversity is not used at all. In the case of microcells this is normal, as diversity provides little gain in microcells where line-of-sight RF paths are

Page 63


predominant. However in large macrocells, lack of diversity can result in significant performance problems.

5.10 BSS Operations Review The following aspects of BSS operations should be reviewed:

5.10.1 Frequently Occurring Alarms A high frequency of occurrence of a particular alarm of group of alarms suggests a specific software or hardware problem in the network (usually vendor-specific). Sometimes it indicates a deficiency in a routine maintenance procedure (eg. frame slip alarms indicating poor calibration of BTS clocks).

5.10.2 Frequency of Outages Frequent BTS or BSC outages cause a significant impact on network quality, due to loss of coverage, dropped calls, interference, and so on. This should be taken into account when considering network quality, alongside network performance statistics.

5.10.3 Transmit Power Calibration It is important that all BTS radios are properly calibrated within the defined range, to ensure the proper calculation of handover and power control algorithms.

5.10.4 External Alarms External alarms are not always provided, but are strongly recommended. They warn about environmental failures such as power failure, air conditioning failure, intruder access, and so on.

5.10.5 Maintenance Schedules It is advisable to check that regular maintenance is carried out according to manufacturers recommendations. This can include the following: •

Radio Calibration

•

Clock Calibration

•

Antenna VSWR Checks

•

Equipment filter cleaning

•

Earth testing

•

…and so on.

The lack of proper regular maintenance results in poor equipment performance and a high rate of failure, all of which contributes to poor network quality. Page 64

Guideline for Network Design and Optimization

Recommend Documents