Using RF optimization tools improves wireless network reliability and performance -

Using RF optimization tools improves wireless network reliability and performance

The introduction of complex data services coupled with the immensegrowth in voice services have placed critical performance demands on wireless accessnetworks.

This situation is complicated by the need for 'all-you can-eat'pricing to attract existing customers to perform trial on new offeringsand to draw customers away from the competition. Wireless accessnetworks continue to hinder the delivery of these capabilities to themarket.

To address this challenge, radio technology standards are evolvingrapidly. Service providers offering cdma2000 have introduced EV-DO technology and are lookingbeyond, to the next technology over the horizon.

The rapid evolution of network technology greatly benefitsend-users, but at the expense of service providers who must makesignificant and frequent capital investments to adopt these solutions.

In an environment with such frequent churn of infrastructurehardware and software, focusing on the optimal use of existing networkinfrastructure and minimizing the purchase of additional components arecrucial.

Service providers who fail to track the service levels of theircompetitors may rapidly lose customers to their competition. Those whomeet customer needs by over-engineering suffer significant penaltiesdue to excessive capital expenditure and are not able to keep theper-user costs down.

Figure1: RF optimization solutions can improve call drop rates and delay theneed for carrier upgrades per period.

Frequent RF optimization is required to best utilize embeddednetworks, enabling them to provide high QoS with minimum capitalinvestment. RF optimization solutions can provide several benefits (Figure 1 above ).After thedeployment of new RF power settings, voice quality is improved and calldrop rates are lowered. Meanwhile, the load on the network achievesbetter balance after RF optimization, delaying the need for carrierupgrades.

Network optimization
A proposed solution for RF network performance optimization is shown inFigure 2, below.

Figure2: Coupling internal software tools and customer-specific interfacesoffers a comprehensive solution for radio network optimization withoutrequiring drive tests or cell site visits.

A combination of internal software tools and customer-specificinterfaces offers a comprehensive solution for the optimization of aradio network without requiring drive tests or cell site visits. Thissolution includes the following components:

Networkconfiguration data . Several inputs to the simulation model arerequired to accurately represent a live network. Network configurationdata is used to track the current equipment deployed in the field. Bysetting interfaces that directly read the outputs from networkmanagement systems, up-to-date network configuration data can bemaintained.

Likewise, the data can be crossed check to identify areas where itmay not be consistent. Network configuration data includes networksettings that are not specific to a particular cell site.

These include maximum mobile transmit power, antenna relatedinformation, specific information on cell sites (such as the number ofsectors channel elements installed), specific information on sectors(such as the number of carriers and their location) and specificinformation on carriers (such as overhead power settings).

Traffic data. Whilesome solutions can operate from long-term averages that can range to amonth, changing geospatial demands placed by highly mobile usersrequire improved granularity of analysis.

The proposed solution retrieves traffic data on an hourly basis suchthat it can be tailored to allow use of traffic data down to thehour-by-hour level for environments in which higher levels ofaggregation produce ineffective solutions.

Simulationmodel. Using hourly traffic inputs and daily networkconfiguration data inputs, the simulation model mimics a live networkas closely as possible. The detailed software simulation modelfeatures:

1) Integration of forwardand reverse link models into a combined simulation model;

2) Vendor-specific carrierassignment; and

3) Monte-Carlo techniques whichinstantiate users and consider separately the SIR requirements, trafficmodels and level of mobility for each user type.

The simulation core allows the analysis to proceed at a level offidelity not possible when using general analytical and heuristicapproaches.

Calibration. It is necessary to calibrate the simulation model against the actualnetwork to ensure that the simulation's response to network changesaccurately reflects the response of an actual live network.

This process is time consuming and requires significant case-by-caseinsight into the workings of the actual network and the simulationmodel. The solution discussed here provides automatic calibration inwhich the simulation adapts itself to more closely resemble the actualnetwork.

Simulated metrics are compared against actual measurements from thenetwork. If the measurements do not match within a prescribedtolerance, an iterative adaptive calibration procedure is used toperturb the probabilistic model of network usage.

Optimization. CDMA-based systems such as the current cdma2000 and UMTS networks enable theadjustment of many parameters, but service providers often adhere tothe manufacturer's default settings.

This is because manual modification of these parameters can producepoor network performance, given the complex, non-linearinterrelationship between these parameters.

The proposed solution computes needed adjustments in RF overheadpower to balance the network load in a way that increases networktraffic-handling capacity.

It does so by shifting the network load from heavily loaded sectorsto lightly loaded ones, ensuring that coverage holes are not created inthe process. Similar optimization algorithms can also be applied totilt optimization when the network has remote tilt antennas.

Prediction. The proposed solution provides predictive graphical and tabular reportsfor the set of recommended RF overhead power adjustments provided bythe optimization algorithms.

Graphical reports aid in providing visual insight into theconsequences of the changes in terms of coverage and serviceavailability. These graphical reports include a best server plot thatshows the strongest pilot signal level Ec ,another best server plot that describes the chip energy to noise andinterference power spectral density ratio Ec /Io aswell as an estimate of locations with available voice services.

Tabular reports are used to inspect the proposed changes before theyare automatically uploaded into the network. There is a report for eachsector, which includes a list of the network parameters extractedautomatically from the network management system and used as input tothe simulation model.

There is also another report for each sector in which overheadchannel power modifications are recommended. The latter report includesthe values of the pilot, paging and sync channels as they currentlyexist and as they should be set after deployment.

Deployment .Recommended RF overhead power adjustments can be deployedinstantaneously in a network via Deploy and Restore scripts, thussaving the cost of sending field crews to the site.

Employing the proposed solution allows these changes to be easilyuploaded into the network, enabling the changes to be made morefrequently than before. This allows designers to perform networkoptimization in near realtime to meet temporal changes in thegeographic distribution of customers.

Data collection. After the deployment of recommended RF power adjustments, real-timedata is collected on the network to verify the results and responsespredicted in the simulation model.

Several key performance metrics are monitored for this. Using thisdata, designers can track the period when new RF optimizationrecommendations are needed. The system also learns from this data suchthat it can perform better in the next implementation.

Verification. Theeffectiveness of the recommended RF overhead power adjustments can beanalyzed after several weeks of deployment, in which sufficient datahas been collected.

The proposed solution includes several tabular reports. One reportwill show the pilot channel power and average power loading before andafter the deployment for each sector in which overhead channel powermodifications were made and for relevant sectors in the vicinity ofthose sectors.

Another report will tabulate key performance metrics both in thepre- and post deployment periods for each sector in which overheadchannel power modifications were made and for relevant sectors in thevicinity of those sectors.

Figure3a and 3b: Heavily loaded sectors were surrounded by lightly loadedsectors, requiring only one or two carriers (a). Overhead poweradjustments were recommended for the region to grow certain sectorswhile shrinking others (b).

Case study
Using the proposed solution, a trial was conducted for a serviceprovider in a market where several sectors were near carrier exhaust,already using three carriers.

A fourth carrier was already planned for the region. In Figure 3a above , heavily loadedsectors were surrounded by lightly loaded sectors, requiring only oneor two carriers. Several overhead power adjustments were recommendedfor the region to grow certain sectors while shrinking others (Figure 3b above ).

Note that a sector is represented as a pie that shrinks or growsproportional to its overhead power setting. In this trial, thedeployment of the proposed RF power changes delayed the need for theadditional fourth carrier by nine months.

Howard Sherry is Chief Scientist,Stephanie Demers and Gregory Pollini are Senior Scientists at Telcordia Technologies Inc. Toread a PDF version of this article go Optimizewireless networks for end-users.  

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.