Why We Made These Wireless Engineering Considerations — Real Projects, Real Reasoning

The best wireless infrastructure decisions aren't always obvious. This article explains the specific engineering reasoning behind design choices we've made on real projects — so you understand what you're buying and why it was built that way.

One of the most common questions we get from clients, procurement committees, and successor IT staff isn't about performance specifications or coverage maps. It's simpler: why?

Why did you use licensed microwave instead of fiber on that corridor? Why did you design a ring topology when a hub-and-spoke would have cost 30 percent less? Why is the power supply rated for twice the operational load? Why did you leave three frequency channels unassigned in the plan?

Wireless Engineering Considerations

These are good questions. Infrastructure that can't explain itself is infrastructure that can't be maintained, extended, or optimized by anyone who wasn't in the original design conversation. This article provides the answers wireless engineering considerations— not as abstract engineering principles, but as the specific reasoning we apply to real design decisions.

Wireless Engineering Considerations: Licensed Microwave Backhaul Instead of Fiber

A municipal utility client needed to connect a new operations center to the primary network hub 6.2 miles away. The apparent solution was fiber. When we analyzed the path, a different picture emerged.

The fiber route required permits for two railroad crossings, a state highway bore, and a floodplain crossing that triggered an environmental review. The estimated timeline: 18 to 24 months, with meaningful permitting risk at each crossing. The construction cost estimate: $1.7 million to $2.1 million depending on permitting outcomes.

We designed an 11 GHz licensed microwave link instead:

•       FCC frequency coordination completed — protected, interference-free spectrum assignment for the specific path

•       Link budget calculated to deliver 99.999% availability using ITU-R P.530 path loss models for the specific climate zone

•       1+1 equipment protection (hot-standby radio) with sub-second automatic failover

•       10 Gbps capacity with headroom for future growth

•       Out-of-band management integrated at both endpoints

Project timeline: 11 weeks from design start to commissioning. Total project cost: $265,000.

The client received a higher-availability, better-documented, faster-deployed connection for approximately 14 cents on the dollar of the fiber alternative. The fiber project was shelved — not because fiber was wrong in principle, but because the specific constraints of this path made licensed microwave the engineering-rational choice.

A design decision isn't good or bad in the abstract. It's good or bad relative to the specific constraints — timeline, cost, right-of-way, availability requirement — of the problem being solved.

Decision: Ring Topology Instead of Hub-and-Spoke

Hub-and-spoke is cheaper to build. Each remote site connects directly to a central hub, minimizing link count, radio count, and configuration complexity. It's the default topology choice when cost is the primary criterion.

It's also a topology with a structural reliability problem: any link failure disconnects every site downstream of it, and hub failure takes down the entire network.

For a water utility client operating SCADA across eleven pump stations distributed across a 40-square-mile service area, hub-and-spoke meant that a single failed backhaul link would take three to four pump stations offline simultaneously — stations whose automated control logic required continuous SCADA communication.

We designed a ring topology: each site connected to two adjacent nodes, forming a closed loop. The design delivers:

•       Single-failure resilience — any link, node, or power event leaves every site connected through the alternate ring path

•       Automatic self-healing using Rapid Spanning Tree Protocol (RSTP), with convergence times under 50 milliseconds — no operator intervention, no truck roll, no manual failover procedure

•       Fault isolation — the network management system identifies exactly which link or node failed, giving technicians a precise diagnosis before they leave the office

•       Graceful capacity growth — new sites can be added as ring nodes without redesigning the topology

The cost delta over hub-and-spoke: approximately 32% additional capital. The first time the ring self-healed around a link failure during a wind event — with zero operator intervention and zero operational impact — the client stopped asking about the cost difference.

Decision: Power Supplies Sized at 70% of Rated Capacity

Equipment manufacturers rate power supply output at the ceiling — the maximum the unit can deliver under controlled conditions. Operating at 100% of rated capacity means operating at peak thermal stress with no margin for environmental variation.

AO Wireless specifies power supplies sized so that the full operational load of the site lands at 70% of rated capacity or below. The engineering rationale:

•       Thermal headroom: Power supply efficiency decreases at high load, and heat is the primary driver of component aging. A unit running at 70% of rated load in a 38°C outdoor enclosure runs significantly cooler than one at 95%, translating directly to longer MTBF

•       Growth accommodation: Networks get expanded. Cameras get added. Sensors get deployed. A power system with 30% headroom absorbs that growth without redesign. A power system running at capacity requires a change order for every addition.

•       Battery interaction: UPS and battery backup systems have better charge and discharge efficiency characteristics at moderate load levels, improving runtime during power events

•       Temperature resilience: Outdoor equipment enclosures see significant ambient temperature variation across seasons. A power supply with thermal headroom in summer is a power supply that doesn't fail in July.

The additional capital cost of specifying a larger power supply is typically $150 to $300 per site. The cost of an emergency power supply replacement at a remote site — parts, labor, travel, and the operational cost of the outage — is typically $3,000 to $8,000. The math is not complicated.

Decision: Leaving 35% of Available Spectrum Unassigned

When developing a channel plan for a multi-site wireless network, we don't fill every available channel. We deliberately leave 30 to 40 percent of the usable spectrum unassigned.

This is counterintuitive — why pay for spectrum capacity you're not using? The reasoning:

•       Interference adaptation: RF environments change. New wireless deployments appear nearby. Interference sources relocate. Vegetation grows. Reserved spectrum gives us the ability to re-plan around new interference sources without disrupting live traffic on the network.

•       Capacity growth: A network operating at 100% of available spectrum capacity has nowhere to go when traffic demand grows. Reserved spectrum is future capacity headroom — available without a frequency redesign.

•       Troubleshooting flexibility: When a link is underperforming, one of the first diagnostic interventions is shifting to an alternate channel. In a fully-assigned plan, that option doesn't exist without recoordinating the entire network. Reserved channels eliminate that constraint.

•       Adjacent channel isolation: Leaving guard bands between assigned channel groups reduces adjacent channel interference and improves the aggregate noise floor across the network — particularly important in dense multi-site deployments.

Spectrum is not a use-it-or-lose-it resource. Unassigned spectrum is a design asset: flexibility, growth headroom, and troubleshooting capability that you'll use before the network's design life is up.

Decision: Physically Diverse Paths for Redundant Links

When a network includes redundant links, physical diversity is what makes the redundancy real. We've assessed networks where the primary and secondary links ran on adjacent poles on the same street segment — a single vehicle collision with a pole took both links offline simultaneously. We've assessed ring topologies where every site in the ring connected to the same mounting structure at the hub — a single structural event would have severed every ring path at once.

Physical diversity means different poles, different conduit runs, different power infrastructure, and ideally different geographic corridors between major nodes. For two-path designs between critical facilities, we route the paths on opposite sides of the building, the block, or the corridor — so the failure modes of the two paths are genuinely independent.

For the highest-criticality links, we also consider equipment diversity: primary and backup links from different manufacturers where the application allows it. Software defects and firmware vulnerabilities don't always affect all vendors simultaneously. A bug that takes down one vendor's radios across a network won't affect a backup path running a different platform.

Redundancy that fails under the same conditions as the primary isn't redundancy. It's an expensive backup that isn't there when you need it.

Decision: Out-of-Band Management at Every Site

Out-of-band (OOB) management provides a separate, independent path to access and control network equipment — a path that remains functional even when the primary data network is down.

Without OOB management, diagnosing a wireless network failure requires physical presence at the affected equipment. For a site on a rural water tower, that's an hour or more of drive time each way, plus the climb. For an urban site with access control, it's coordination with a property manager on top of the drive. For a remote site accessible only by seasonal road, it might mean waiting for conditions to permit access.

With OOB management — typically implemented through cellular modem integration in the equipment enclosure, connected to a dedicated management network — our engineers can:

•       Access the configuration interface of any network device remotely, regardless of primary network status

•       Diagnose whether a problem is hardware, software, configuration, or power — and often resolve it — without leaving the office

•       Push configuration changes, perform firmware updates, and reboot equipment remotely

•       Monitor the site's power system status independently of the primary network link

OOB management turns a 4-hour diagnostic truck roll into a 20-minute remote session. Over the life of a network, that time savings is substantial. We include OOB management in every network we design — it's not optional infrastructure, it's how we make sure we can take care of the networks we build.

The Design Philosophy That Connects These Decisions

Every decision described in this article comes from the same underlying premise: we're building infrastructure that needs to work, not infrastructure that needs to pass an acceptance test.

Passing an acceptance test is easy. You demonstrate link connectivity, verify throughput at a point in time, show the coverage predictions match the measurements. Done.

What happens 18 months later — when the RF environment has shifted, when the power supply is running at 95% in a 40°C enclosure, when someone needs to add three cameras and the power system can't carry them, when a link fails and there's no OOB management path to diagnose it remotely — that's the real acceptance test.

We design for that test. The decisions cost more upfront and require longer design conversations. Our proposals are sometimes more expensive than competing bids.

They also produce networks that work — reliably, transparently, and for the full design life — without the remediation cost that comes from having to do it over.

Frequently Asked Questions: Wireless Infrastructure Engineering Decisions

When is licensed microwave backhaul better than fiber?

Licensed microwave backhaul is often preferable to fiber when the fiber route requires complex permitting (railroad crossings, highway bores, floodplain crossings), when the timeline is shorter than fiber permitting allows, when the per-mile cost of fiber construction exceeds the microwave alternative, or when the link requires carrier-grade availability but the right-of-way risk makes fiber reliability uncertain. Licensed microwave can achieve 99.999% availability at a fraction of the cost and timeline of difficult fiber routes.

What is a ring topology in wireless networking and why is it used?

A ring topology connects each network site to two adjacent nodes, forming a closed loop. If any single link or node fails, traffic automatically routes the opposite direction around the ring, typically converging in under 50 milliseconds using Rapid Spanning Tree Protocol. Ring topologies are used for critical infrastructure — SCADA, public safety, municipal backhaul — where any single link failure cannot be allowed to disconnect sites.

Why are wireless network power supplies oversized?

Power supplies in outdoor wireless infrastructure are specified at 70% or less of rated capacity to provide thermal headroom (extending MTBF significantly), growth accommodation for equipment additions, better interaction with battery backup systems, and resilience during ambient temperature extremes in outdoor enclosures. Operating a power supply at maximum rated capacity in a hot outdoor environment is a common cause of premature failure.

What is out-of-band management for wireless networks?

Out-of-band (OOB) management is a secondary, independent network path — typically a cellular data connection — that allows remote access to wireless network equipment even when the primary data network is down. It allows engineers to diagnose and often resolve wireless network failures remotely, without a truck roll to the physical site. OOB management dramatically reduces mean time to resolution for network incidents.

Why do wireless engineers leave spectrum channels unassigned?

Leaving 30-40% of available spectrum unassigned provides interference adaptation headroom (the ability to re-plan around new interference sources without disrupting live traffic), future capacity growth, troubleshooting flexibility (the ability to move a link to an alternate channel during diagnostics), and guard bands that reduce adjacent channel interference. Available spectrum is a design asset, not unused capacity.

Have a wireless infrastructure project or just want a second opinion? Contact the AO Wireless engineering team at www.aowireless.com. No pressure — just straight talk from people who've built it.

Free Consultation