BACKHAUL DIMENSIONING INPUT

One result of the dimensioning of the access network is the required capacity by the backhaul network. A typical backhaul network consists of point-to-point or point-to-multipoint radio links or even fiber rings in recent networks. In most cases the information rate that is originated to and from the backhaul network should be calculated to evaluate an existing infrastructure or consider the cost to deploy a backhaul network along with WiMAX.

The total committed information rate of a WiMAX PoP is linear to the number of sectors and average sector throughput, which mainly depends on channel bandwidth and deployment scenario. Considering trisector cells, and average throughput of 9 Mbps (5 MHz channel bandwidth), the total TDD committed rate is 27 Mbps. To estimate the FDD equivalent, which is directed to and from the backhaul network, the total rate is multiplied by the DL/(DL + UL) ratio. Hence the traffic that should be reserved in the backhaul network for a PoP with 2/3 DL WiMAX traffic is 18 Mbps. For a dual layer cell or 10 MHz channel bandwidth the backhaul rate per PoP would be doubled, around 36 Mbps. It is acceptable to include a 5 percent margin as in practice it is common to observe small deviations in sector throughput. During the deployment and the operation the exact backhaul traffic can be monitored through the management system and design adjustments can be made accordingly.

In certain occasions the network designers are requested to propose a backhaul system along with the WiMAX network. Although this is not in the scope of this chapter, however there are some interesting observations that can be highlighted. Clearly the multipoint systems are more cost effective than point-to-point radio systems when the WiMAX PoP number is high, the PoP are quite close to each other and when their backhaul rate is such that more than 3–4 WiMAX PoP per multipoint sector can be served (around 10–12 per site). This indicates that the WiMAX site backhaul traffic should be from 15 to 25 Mbps, which mainly refers to trisector cells with 5 MHz channel. For higher bandwidth systems, such as dual layer cells or with 10 MHz bandwidth per site the point-to-point solution may offer the higher required capacity. Again in a dense urban environment where sites are usually deployed with separation of 0.3–0.7 km the use of fiber (if available) could be a better solution. Concluding, the best backhaul system strategy can be determined by evaluating the number of WiMAX PoP, their backhaul rate, and finally the PoP positioning. Note that although the WiMAX network dimensioning is done in such way to optimize the access network, this does not suggest that the backhaul network is also optimized and a separate study may be necessary.

DIMENSIONING PROCESS

The process of dimensioning involves a sequence of steps, which serve different requirements such as capacity or coverage estimations, to conclude to a final outcome as shown in Figure 1. The input consists of the business plan, the assets, and the key performance indicators (KPIs). The first action is then to verify that all necessary information is available and clarified, alternatively an interactive “questions and answers” session is necessary. During the input analysis, the designers utilize their theoretical and technological expertise to define the dimensioning strategy. In somecases this analysis identifies that potential business plan assumptions could affect the network design process, and a discussion is likely to be initiated with the customer. Such an approach is more appropriate after the award of a contract. The following step is the processing of service characteristics such as the Internet rates and Voice-over-Internet Protocol (VoIP) codecs. In most cases the services are provided in the business plan with their marketing description and it is necessary to identify the impact of each service on the WiMAX air-interface. In the next step, the coverage analysis is performed: the provided service areas are identified and the required number of points of presence (PoP) is estimated. A PoP refers to a WiMAX site, with at least three sectors, which is capable of providing a 360° footprint.

 

Figure 1: Network dimensioning process.

Further to the coverage analysis, the service areas should be also offered sufficient capacity as dictated by customer numbers and services profiles. Therefore the next step is to estimate the required number of sectors to achieve this capacity. The estimates of PoP and sectors are then used to determine the configuration of the BS in the final joint analysis step. It is common after the joint analysis for designers to evaluate the results and proceed with customizations to further improve the solution. In this case, coverage, capacity, and joint analysis may be revised several times before providing the final output. A description of the strategy, the final bill of material (BoM), and a discussion of the assumptions and methodology are mainly the outcome of a dimensioning study.

1 BUSINESS PLAN, ASSETS, AND KEY PERFORMANCE INDICATORS

A well-defined business plan can be the most significant driver behind a successful investment on WiMAX technology. From a cost point of view, a very ambitious plan in terms of subscribers/services will certainly impact CapEX and OpEX, whereas from a design point of view, the impact comes when the requirements are not extensively defined or they contradict the technology capabilities. A revision of the business plan during or after the network design may compromise the network performance or create additional design costs but most importantly it will result in implementation delays. A list with the most common information provided with a business plan is presented as follows:

1. Service area(s): defined with geocoded polygons, including the size in km², and the terrain profile details (i.e., urban, suburban, rural, average building height, etc.).
2. Coverage type: such as fixed-outdoor, on rooftop, or on outer walls, fixed-indoor, nomadic outdoor/indoor, mobile outdoor or any combination thereof.
3. Subscriber profile(s): such as residential, small business, corporate. Subscriber profiles may relate to a specific type of coverage and service.
4. Subscriber distribution: subscriber numbers per profile, per service area, and per deployment year, according to the scalability plan.
5. Service profile(s): such as VoIP, broadband Internet, VPN along with their distinct characteristics (i.e., VoIP codecs, peak information rates, contention factors, etc.). Service profiles may relate to specific subscriber profiles and coverage types.
6. Available spectrum: defined as paired, along with local regulations concerning the allowed channelization and duplex schemes.
7. Existing infrastructure: such as sites that can be reused, available backhauling equipment with Ethernet interface, and core network PoPs.
8. Cartographic data: such as high-resolution digital maps with buildings.
9. Key performance indicators: such as coverage objective in terms of percentage of the service area, differentiated per terminal type, where a stable QPSK link can be achieved.
10. Customer requirement: such as duplex scheme, number of sectors/BS, channel bandwidth, reuse scheme, type of sites, deployment strategy.

2 STRATEGY

During request for information (RFI)/RFP stages, a dimensioning exercise may be requested by a customer, mainly for two reasons: either to acquire know-how by differentiated proposals or to identify the more cost-efficient solution. In the first case, the requirements are usually relaxed so that the participant vendors/integrators can design with flexibility, while the provided information (i.e., business plan, assets, service areas) is hypothetical. The submitted studies will probably be presented in various formats and most certainly based on diverse assumptions. In such case a direct comparison among the studies is complicated, and usually a more defined exercise is the next step. In the second approach, the case study is well defined so that the design assumptions are either implied or directly mentioned. The results are now directly comparable, hence a clear ranking list can be obtained. From RF network designer point of view a different strategy should be followed: showing flexibility in the network design and perhaps providing several alternatives for the first approach, while a more strict, cost-optimum solution is more appropriate for the second approach.