SCALABILITY | STRATEGY INTERACTION

As mentioned previously, the most significant advantage of WiMAX is the flexibility and scalability of the air-interface. Therefore the concept “pay as you grow” is definitely applicable in WiMAX commercial networks. Scalability allows the reduction of initial investment and risks, and thereafter the network is expanded based on the market penetration and revenues. The majority of business plans provide the subscriber numbers and services per deployment year and in this context the dimensioning should be presented according to the scalability plan (network size/performance per year). The complexity in this case is not the increase of subscriber numbers or service area size, but when in parallel the coverage type is upgraded or new products (based on recently released standards) have to coexist with deployed equipment. The main types of scalability plan that have to be considered during dimensioning are presented as follows:

• Network expansion: Expansion may be in terms of subscriber numbers, service rates, or service areas or a combination and involves the deployment of new PoP or addition of sectors in existing PoP. A great challenge, in such scenario, is to optimize the positioning of PoP to achieve the best coverage and capacity outcome. It is more appropriate to determine the network size and PoP positions for the last deployment year, and then deploy only a subset of PoP that would satisfy the objectives of the initial phase.
• Coverage upgrade: It is common to allow the use of more demanding terminal profiles (i.e., nomadic/mobile) in the network after the first year so as to allow time to test the performance of the wireless network. In this case, not only the network is expanded from first to second year, but also the coverage should be upgraded too. The same approach “design for the future, deploy for present” as above should be applied, and the only difference is that a more dense network is probably required.
• Technology upgrade: The major change in terms of equipment is between the IEEE 801.16-2004 and 802.16e standards. The new products are based on software defined radio technology and therefore future standard releases will probably be implemented with minor changes. It is quite challenging to upgrade an existing fixed WiMAX network to coexist with mobile WiMAX, unless there is provision for additional spectrum. The major challenge is to replace subscriber equipment and restore access and it is likely that this transition phase will take a long time.

CUSTOMIZED NETWORK DESIGN | WiMAX Networks Dimensioning

Customized network dimensioning provides a major advantage to vendors/integrators by highlighting their expertise and by indicating a cost-optimum solution. In many RFPs or project cases the developed business plan has extensive details which can be exploited in a very positive manner. A common occasion is that the overall service area is broken down into subareas with distinct characteristics, such as common type of customers and terminal profiles, common terrain, or service requirements. Such distinction allows a customized treatment of each subarea, where its characteristics are matched with an optimum solution, thus avoiding an overall rough approximation. The main customizations that can be applied in dimensioning, provided that the necessary information is available, are described below:

• Nonuniform sectorization: The use of different sectorization schemes, depending on the coverage type may result in reducing the required PoP. In some subareas there might be only SMEs, where a fixed-outdoor unit would be utilized. Therefore, in this case there is no restriction to use trisector cells. If the available spectrum is sufficient, up to eight sectors can be deployed in a PoP, depending on the capacity requirements.
• Dual layer coverage: In many occasions the subscribers will use both fixed-outdoor and mobile units hence in this case a more efficient approach is necessary. If extreme capacity is required, a high number of sectors can be used, building a trisector layer for the mobiles and overlapping layer for the fixed, provided that frequency reuse can be applied. The neighboring list for mobile handovers will include only a sector in the first layer, while the demanding fixed links will be isolated in the second layer. This approach is very cost effective in cases where an extreme capacity demand would require a PoP every 0.2–0.5 km, if plain trisector cell layout was considered.
• Nonuniform channel bandwidth: Assuming that extreme capacity is still the performance indicator, by using higher channel bandwidth the capacity per PoP is increased and therefore the number of PoP is maintained at a reasonable level. While the overall network may be implemented with trisector cells, 5 MHz channel bandwidth and 1.3.3 reuse scheme (using 3 out of 4 available channels), a specific subarea with SMEs may be served by quad-sector cells, with 10 MHz bandwidth. In this case, the 10 MHz in the upper band will be assigned in sectors 1 and 3, while the 10 MHz in the lower band in sectors 2 and 4. This arrangement provides a 260 percent increase in capacity for the same footprint. It should be noted that the use of nonuniform channel bandwidth requires specialized sector configuration (transmit power, UL receive target level, and antenna arrangement) and extensive interference studies, applying the interference rejection filter methodology that takes into account the impact from trasmissions in all channels.
• Hybrid coverage: When both fixed-outdoor and nomadic/mobile coverage is required, the design approach is to select the operating range for the worst case condition. This usually leads to a huge upfront network size and investment. An alternative approach would be to consider that only a portion of the service area will be covered with the worst case terminal, hence the cell range can be selected higher. As the network evolves and additional PoP will be deployed for capacity upgrade, the percentage of the mobile coverage in the service area will also increase. This approach is particularly efficient for networks that extend to a large area, however with low subscriber density. The dimensioning is based on fixed-outdoor coverage that can reach higher ranges, however close to a BS there will be opportunity to use fixed-indoor/nomadic/mobile terminals also. 
• Divide and conquer principle: To apply the previous customizations, the service area can be processed in pieces, where each subarea has a specific characteristic. If the segmentation of the service area is not available in the business plan, the designers can perform such action, by requesting or collecting more information (i.e., from site survey). This principle usually drives the network size up, and it is most useful for the actual design phase. The equipment quantities are increased in an exact manner because greater detail is taken into account during the actual design. This method approaches an optimum wireless network design.

It should be mentioned that the advances of the WiMAX air-interface allows greater flexibility and customizations during the wireless network design and hence balances the complexity of accommodating terminals with different profiles.

PRESENT–FUTURE TECHNOLOGIES

Over the last two years IEEE 802.16-2004 products have become more widely used and with several deployments in the field, basic experience with WiMAX technology has been increased. The new, upcoming IEEE 802.16e, which is an amendment to the IEEE 802.16-2004 standard, promises significant improvements. The major enhancements in WiMAX technology for the upcoming version of the standard are outlined in Table 1.

Table 1: Evolution of IEEE 802.16 Technology 

Technology	IEEE 802.16d	IEEE 802.16e	IEEE 802.16j
Range (km)	0.7	1.8	2 (per hop)
AAS	–	STC/MRC/SM/BF	Cooperation
Air-interface	OFDM	OFDMA	Relay-based
QoS	Basic	E-rtPS	Enhanced
Profile	IP Padios	ASN	ASN
Scenarios	Nomadic	Mobile	Mobile relaying

The most important enhancement concerns the system gain, which is roughly increased by 15–25 dB, and hence the cell range is also increased. For fixed/nomadic terminals it is evident that 802.16e is much more efficient and is also capable of catering for mobile terminals. Another decisive improvement is the introduction of AAS, which increases robustness through STC, MRC, and spectral efficiency through spatial multiplexing (SM) and BF. The transition to OFDMA clearly boosts the radio resource management efficiency and improves QoS particularly in the presence of VoIP service. It is evident that the evolution of WiMAX targets three objectives: to increase the system gain and reach customers inside their homes/offices and on the move, to boost capacity so as to reduce service costs and be competitive with other access technologies, and finally to coexist in a seamless manner in the upcoming all IP networks.

This is clearly the case with the development IEEE 802.16j which introduces the concept of relays. A relay-based network can in principle extend the range boundlessly, however, in practice 2-hop links are more likely to be implemented (for delay and throughput issues). From the designer’s perspective if the first link (BS-relay) is line of sight then the system range can be several times higher than conventional systems, and hence for coverage-limited networks the dimensioning would result in much reduced costs. Furthermore, if the BS employs BF toward the relays, concurrent communication with several of them may be established in the form of spatial-division multiple access (SDMA), therefore boosting cell capacity.

While the all IP architecture is on the way (access service network (ASN) gateway), with major manufacturers of network products supporting this direction, the next WiMAX standard, IEEE 802.16m is also under consideration. IEEE 802.16m will revise the air-interface in the scope of international telecommunications union–radiocommunication sector (ITU-R) requirements for IMT-2000 and IMT-Advanced.

DIMENSIONING CERTAINTY AND MARGINS

As stated above, dimensioning accuracy is crucial. For vendors/integrators, it is essential that an RFP response should be financially prudent to increase chances for a contract award. However after the award, the network should be deployable, and considering that the majority of RFPs concern turn-key projects, a rough or underestimated offer can lead to miscalculation of the required network size and hence increase the implementation costs at the vendor’s/integrator’s expense. A balanced condition can be achieved by incorporating operating/performance and certainty margins. An operating margin is applied to coverage estimations, where the system range is selected smaller than the maximum. The same applies to capacity estimations where, for the average sector throughput, usually a small margin is considered. Considering the overlapping effect of various margins, a careless consideration can lead to overestimation. Specifically for coverage and capacity margins, the network will be either coverage-limited or capacity-limited. Therefore, in practice only one of the previously mentioned margins has impact on the financial offer. An important factor for defining margins is the quality of the provided business plan. The provision of extensive information would facilitate more accurate dimensioning.

It should be noted that in addition to the operating margins, another issue to consider is the network implementation margins. Although the RF designers take great caution to predict any possible causes of degradation, in many occasions problems may occur during the implementation. An engineering team, which is not well trained, or pays little attention to details, can make the difference between a successful and poor deployment. In cases where existing infrastructure is utilized, such as sharing of GSM sites, the condition of these sites or the restrictions posed by an operator usually cause problems. For example to save cost of antenna poles, operators may install WiMAX antennas below GSM antennas. This is contradictory to best practice since WiMAX operates in higher band and hence experiences higher propagation losses, and furthermore there is the issue of equipment RF isolation where a minimum separation distance should be maintained. Another example is with the installation of fixed-outdoor CPEs. Careless installation will result in suboptimum performance of such units, compromising their competitive advantage which is high capacity and robustness. In general, past deployment experience can make the difference in dimensioning, and it is preferred that the RF network designers which are involved in presales activities are the same that will have the responsibility of carrying out the final design and deployment supervision.

EQUIPMENT SPECIFICATIONS

From a WiMAX access network design perspective, the most important parameters are related to the PHY and MAC characteristics of the air-interface. Typically all parameters that affect the link gain budget, such as transmit power, antenna gains, receiver sensitivities, advanced antenna systems are of great importance. Consider an example of a WiMAX access network that is intended to serve stationary indoor terminals in a typical suburban environment (i.e., SUI-C channel), hence a coverage-limited scenario. Comparing two systems with MIMO 2x and BF 8x configuration,the system range is around 2.2 and 3.7 km, respectively. This result is for the same amplifier output where the 6 dB system gain is due to the difference in antenna elements. Note that the diversity or BF gains are not considered for the range estimations since they are only applied in user traffic and not in the signaling part of the frame that usually limits the range. Applying Equation 13.5, the footprint is estimated as 12.5 and 35.5 km², respectively. Results indicate that for coverage limited scenarios the higher system gain of a BF configuration can significantly reduce the network size. On the other hand, for mobility scenarios the use of MIMO is a more suitable approach. In general WiMAX has several air-interface profiles that may be best suited according to the deployment scenario.

In addition to the link gain budget, the sector capacity is equally significant for capacity-limited scenarios. A comparison of the spectral efficiency per modulation and the required SINR thresholds could indicate that some products may operate with higher efficiency than others. Capacity is further increased by advanced antenna systems, where spatial multiplexing could even double the spectral efficiency, while BF could reduce interference levels and upgrade the PHY mode.

A list of important parameters, with impact on coverage, capacity, and QoS are shown in Figure 1. These parameters should be provided for all combinations of operating frequency band-channel bandwidth, both for the DL and UL and for different terminal profiles. A vital parameter for dimensioning is the number of subscribers that can be supported in a sector due to availability of service flows. This number depends on the number of service types per user. Additionally, a description of the capabilities and performance of the radio resource management (scheduler) would be beneficial to the designers. It is common during dimensioning to assume that the system can preserve QoS but in many occasions a safety margin (i.e., 5 percent) is required. The scheduler’s performance may downgrade close to full capacity load or for high number of subscribers per sector. Considering BE traffic, the impact is not as significant, however, this is not the case for VoIP and other delay-sensitive services. When handover is involved, it becomes more challenging to preserve QoS. It is evident from this section that WiMAX networks designers should have a thorough under-standing of technology so as to foresee potential issues during network dimensioning and hence consider appropriate margins. A good insight into WiMAX air-interface performance issues.

 

Figure 1: Important WiMAX air-interface parameters.