CAPEX Savings with Antenna Tilt-Based Load Balancing SON Abstract The principal benefits of Self-Optimizing Networks (SON) are reduced OPEX and CAPEX by both minimizing human involvement in network operation and delaying hardware deployment via optimized network performance and efficiency. While OPEX savings may be difficult to quantify due to intangible labor saving effects of particular SON features, CAPEX savings estimates due to capacity enhancing SON techniques such as antenna tilt-based load balancing can be quite concrete at least in terms of the number of cell sites saved over the life cycle of a network if not in actual dollar amounts. In this paper, we show how the Coverage and Capacity Optimization feature of Load Balancing results in a higher overall capacity in the cellular network, which is equivalent to a higher overall spectral efficiency realization. Increased spectral efficiency results in the need for fewer sites to meet the capacity demands in a network culminating in reduced CAPEX. We present the capacity enhancement results from a simulation using data from a live network as well as examples of savings in cell site deployment obtained through the WiROI business case analysis tool. Reverb Networks, 2012
1. Introduction Self-organizing Networks (SON) have attracted much attention during recent years. In today s networks, system deployment and operations are still largely manual. Moreover, the ever increasing data demand will lead to substantial growth in network capacity requirements and complexity. The automation of such processes will reduce expenses and avoid human errors and data inconsistencies. The introduction of SON in future high capacity cellular networks like LTE will improve network performance and efficiency through more frequent optimization and overall optimal solutions. The technical strengths and benefits of SON have been described in prior white papers. Here we focus on the SON benefits of reducing capital expenditures (CAPEX) through antenna tilt based load balancing solutions. 2. Saving Overview Due to the explosive growth in data traffic without comparable increase in revenue, operators are urged to cut back on their network spending. One obvious driver for the introduction of SON is the substantial reduction of OPEX and CAPEX. For example, the self-configuration SON feature automates these procedures upon reset or first installation of the base stations. The minimization of human involvement in the process results in a significant OPEX reduction. Another example is for network optimization. According to specific optimization goals, SON functions are implemented to respond promptly and automatically to the dynamic changes in networks. Manual efforts and hence OPEX will be reduced, as automated SON functions require reduced involvement in monitoring and updating. The network performance will be improved as well due to the reduced optimization cycles and higher optimal solutions, which means that we can achieve the same performance goal (e.g. capacity and throughput) via less network resources (e.g. number of cells). In other words, CAPEX will be reduced due to the delay in network investments that would otherwise be required without the SON feature. In the following sections we focus on CAPEX reduction by efficient usage of radio resources. First, we demonstrate through a simulation how load balancing through antenna tilt optimization increases the average throughput of a cluster using sample data from a live network. The increase in throughput is then related to an increase in the realized spectral efficiency of the network leading to a reduction in the number of sites required to meet the data capacity demands in a network. The WiROI business case analysis tool is used to simulate the number of sites required with and without the load-balancing SON feature in the network using a variety of cellular network deployments ranging in size and terrain attributes. Reverb Networks, Inc. 1
3. Antenna Tilt-based Load Balancing SON Increases Throughput Reverb Networks antenna tilt-based load balancing optimization is a SON feature that comes under the Capacity and Coverage Optimization as cited in 4G America s SON Whitepaper 1. This load balancing SON feature aims to distribute the traffic more optimally to relieve congestion and utilize resources evenly across the network by optimizing tilts according to the traffic distribution. This optimization generally results in an increased capacity in the network, with the measure of increase being a function of the degree of congestion being relieved and capacity for a data network being the cell or cluster throughput. This section summarizes the findings that were obtained as a result of a network planning tool study performed on a critical zone of a live network to quantify the effects of tilt optimization on cluster capacity. The critical zone consists of a critical cell, defined as a cell that has considerable congestion problem, and its first tier as well as some second tier neighbors. A high ratio of the data users in the critical cell are exposed to delays due to congestion in the early night hours, approximately 9pm to 12pm. The voice users, on the other hand, do not suffer from significant congestion at any time during the day. The target of this study was to quantify the gains that the Load Balancing algorithm can provide in the critical zone. As shown in Figures 1 and 2, respectively, the critical cell in the critical zone is Cell 13412. Two neighbors that might serve the purpose of taking off some load from the critical cell are Cell 13533 and Cell 12981. The figures indicate the coverage of the cells with initial tilt configuration and their corresponding signal propagations. 1 Self-Optimizing Networks: Benefits of SON in LTE, LTE Americas, July 2011. Reverb Networks, Inc. 2
Figure 1 Initial Coverage by Transmitter Figure 2 Initial Coverage by Signal The optimum tilt values that the Load Balancing feature could reach were found by performing an exhaustive search algorithm. The search results indicated that down-tilting the critical cell antenna by 3 degrees and up-tilting Cell 13533 s antenna by 1 degree yields the best performance. No change is Reverb Networks, Inc. 3
required on the antenna of Cell 12981. The coverage of the cells after applying the final tilt configuration and the signal propagation map are shown in Figures 3 and 4, respectively. Figure 3 Final coverage by transmitter Figure 4 Final coverage by signal Reverb Networks, Inc. 4
The potential performance of the considered tilt change was measured by checking the total throughput in the critical cell only and the critical cell plus the best neighbors. The throughput values obtained in network planning tool simulations based on the initial and final tilt configurations shows a 31% increase in the critical cell throughput and greater than 16% increase in the zone consisting of the critical cell and its neighbors. While the above analysis was done to calculate the increase in data throughput as a result of implementing the Load Balancing feature, the same analysis can be applied to voice networks with similar results for Erlang improvement. 4. Capacity, Spectral Efficiency and CAPEX Spectral efficiency in data networks is defined as the throughput per unit area of coverage per unit of spectrum, or for example bits/second/km 2 /Hz. Similarly, the spectral efficiency in voice networks may be defined as Erlangs/km 2 /Hz. The capacity in a voice or data network may then be defined as: Cell Capacity = Spectral Efficiency x Area of the Cell x Total Bandwidth Therefore, by increasing the throughput or Erlangs in a given area through optimal distribution of users and utilization of resources, the Load Balancing procedure described in Section 3 increases the realized spectral efficiency and hence the capacity of the network. Spectral efficiency of wireless data networks is used to dimension the capacity of cellular networks. While initial deployment of cells is determined by the coverage requirements, subsequent addition of cells after addition of carriers on the original cells is exhausted are driven by the capacity demands in terms of both the number of users and the throughput per user. The higher the spectral efficiency of the deployment, as determined by the multiple access protocol such as OFDM as well as capacity enhancing solutions such as Reverb Networks Load Balancing SON, the lower will be the requirement for additional new cell sites to meet the increasing capacity demands. This results in reduced CAPEX as fewer sites need to be added over the life cycle of a wireless data network. While the qualitative relationship between spectral efficiency and CAPEX is straightforward as defined above, the quantitative calculations are quite complex and require capacity planning tools with traffic modeling and demand characterization as well as network planning tools using infrastructure specifications and terrain data. In this paper we have used the WiROI business case analysis tool developed by Wireless 2020 that incorporates capacity and network dimensioning tools as well as the quantitative benefit of the Load Balancing feature on the spectral efficiency of the network to determine the savings in cell deployment over a period of time for different sized networks with different terrain and morphological characteristics. Reverb Networks, Inc. 5
The WiROI tool models a wireless network s traffic and determines the number of sites needed to meet the traffic demand using network attributes such as spectral efficiency and terrain information. As shown in Figure 5, by doing this analysis with Reverb Networks Load Balancing feature on and off, the savings in the number of cell sites can be calculated for up to 10 years. The quantitative benefit of Reverb s Load Balancing feature can be calibrated by using the Optimization Improvement slider shown in Figure 6. This percentage improvement applies directly to the spectral efficiency. Using the inputs for base station cost, the cumulative CAPEX of up to 10 years may also be calculated. In the section 5, we present the results of the number of sites saved by using Reverb s Load Balancing SON feature over 10 years for different size and types of market and varying percentage of spectral efficiency improvement from the SON feature. Figure 5 WiROI screen showing Number of Cells required over 10 years Reverb Networks, Inc. 6
Figure 6 WiROI screen showing Cumulative CAPEX over 10 years 5. Summary of Cell Site Saving with Reverb s Load Balancing SON We used the WiROI business case analysis tool to also calculate the savings in cell site deployment over 5 years due to the capacity enhancing benefits of Reverb s Load Balancing SON feature. The savings were calculated for spectral efficiency improvements of 10%, 15% and 20%, consistent with the type of improvement seen in the simulation described in section 3. In Figures 7 and 8, the three networks chosen for this case study are dimensioned in terms of number of users and deployment of cell sites without any SON feature over 5 years. The three sizes of networks chosen represent small, medium and large network deployments. Reverb Networks, Inc. 7
Subscriber Growth Over 5 Years Number of Subscribers 1000000 900000 800000 700000 600000 500000 400000 300000 200000 100000 0 1 2 3 4 5 Small Network Medium Network Large Network Year Figure 7 Number of Users over 5 Years Figure 8 Number of Cell Sites without Load Balancing Reverb Networks, Inc. 8
In Figure 9, the results of WiROI analysis for three sample markets with and without Load Balancing SON feature benefits are shown. In Figure 10, the CAPEX savings resulting from the fewer cell sites deployed are calculated using $48K as the cost of a standard 3-sector BTS inclusive of cables and antennas. 2 Number of Sites 740 720 700 680 660 640 620 600 580 560 540 Small Network: Sites Deployed in 5 Yrs Number of Sites Medium Network: Sites Deployed in 5 Yrs 950 900 850 800 750 700 650 Number of Sites 1700 1650 1600 1550 1500 1450 1400 1350 1300 1250 1200 Large Network: Sites Deployed in 5 Yrs Figure 9 Number of Cell Sites with and without Load Balancing Small Network: Cumulative BTS CAPEX @ 5 Yrs Medium Network: Cumulative BTS CAPEX @ 5 Yrs Large Network: Cumulative BTS CAPEX @ 5 Yrs $K 34000 32000 30000 28000 26000 24000 $K 44000 42000 40000 38000 36000 34000 32000 30000 $K 80000 75000 70000 65000 60000 55000 Figure 10 Cumulative BTS CAPEX in 5 years In conclusion, these results show a savings in cell site deployment, and hence BTS CAPEX, ranging from 9% to over 16%. WiROI is a trademark of Wireless 20/20, LLC 2 The Wireless Industry's Best Kept Secret: The Price of a Base Station: http://mobilesociety.typepad.com/mobile_life/2007/02/the_wireless_in.html Reverb Networks, Inc. 9