WISP Backhaul for Businesses

A wireless ISP's network has two halves. The access network is the part your subscribers see: the sector antennas on the tower, the radios on their roofs, the unlicensed or lightly licensed spectrum between them. The backhaul is everything behind the tower — the connection that aggregates all of that subscriber traffic and carries it to a point where it can reach the internet, whether that's a carrier hotel, a data center meet-me room, or a provider's regional aggregation node.

Backhaul is sometimes called the 'middle mile,' and the name fits: it sits between your last-mile wireless links and the backbone of the internet. It's the least visible part of a WISP and the most consequential. Subscribers never think about it when it works, and it's the first thing they feel when it doesn't — every speed test, every video call, every stream on every subscriber's device is ultimately bounded by the capacity and quality of the backhaul feeding their tower.

The three main ways to build it are fiber (leased Ethernet or internet access from a carrier, or dark fiber you light yourself), licensed point-to-point microwave (your own radio link on coordinated, protected frequencies), and unlicensed point-to-point radios (cheaper, faster to deploy, but sharing spectrum with everyone else). Most mature WISPs use some combination: fiber where it's economically reachable, licensed microwave where it isn't, and a second path of either type so one failure doesn't darken a tower.

This page is about buying that middle mile well — what the options actually are, what they cost and why, how long they take, and where operators most often get it wrong.

Traffic aggregation: why the tower pipe is smaller than the sum of your plans

A tower with 200 subscribers on 100 Mbps plans is not carrying 20 Gbps. Real networks rely on statistical multiplexing: most subscribers use little or nothing most of the time, and peak simultaneous demand is a fraction of the sold capacity. Backhaul sizing works from the measured peak — typically the evening busy hour — plus headroom for growth. Operators commonly run contention ratios anywhere from 10:1 to 40:1 depending on their subscriber mix and how much congestion they're willing to tolerate at peak. The discipline is in measuring your actual busy-hour utilization per tower and upgrading before the ratio, not the marketing, starts driving the experience.

Fiber backhaul: the gold standard when you can get it

Leased fiber Ethernet — typically carrier Ethernet or dedicated internet access handed off at 1 Gbps or 10 Gbps — offers the highest capacity, the lowest and most consistent latency, and contractual SLAs with defined repair times. If a carrier's fiber already passes your tower or aggregation site, this is almost always the right answer. The catch is geography: towers are deliberately placed where people and infrastructure are sparse, and the cost of extending fiber the last few miles to a hilltop can exceed the tower's revenue for years. That gap between 'fiber is nearby' and 'fiber is at my tower' is where most WISP backhaul decisions actually get made.

Licensed microwave: carrier-class wireless for the middle mile

Licensed point-to-point microwave uses coordinated frequencies (commonly in the 6–42 GHz range, depending on region and path length) that you register and that others can't legally interfere with. A properly engineered licensed link delivers fiber-like behavior — high availability, predictable latency, gigabit-class capacity — across paths of a few miles to a few tens of miles, with availability engineering (rain fade margins, antenna sizing, sometimes space or frequency diversity) tuned to how many nines you need. It requires spectrum coordination, FCC licensing or registration in the relevant bands, and line-of-sight path engineering, but it lets a WISP own its middle mile instead of renting it — often the only economical way to reach a rural tower at real capacity.

Unlicensed point-to-point: fast and cheap, with an asterisk

The same 5 GHz and 60 GHz-class radios used for access can backhaul a small tower over short hops at a fraction of the cost of licensed gear or fiber. For a young WISP's first towers, that's often the correct call. The asterisk is interference: unlicensed spectrum is shared, and a link that tests beautifully at install can degrade as neighbors deploy their own radios, as your own network's self-interference grows, or as seasonal foliage changes the path. Unlicensed backhaul is a reasonable starting point and a poor long-term plan for a tower with meaningful revenue attached.

Redundancy and failover design

Backhaul redundancy means two things that are easy to confuse: a second circuit and a second path. Two fiber circuits from the same provider through the same conduit fail together when the conduit is cut. Real diversity is physical — different routes, ideally different media (fiber primary, microwave backup, or two microwave hops via different intermediate sites) — with routing (commonly BGP or OSPF) configured so traffic moves automatically when the primary degrades. Many WISPs also bring two upstream providers into their core so a provider-level outage, not just a path-level one, doesn't take the network down.

• Peak-hour congestion: evening utilization climbs, latency spikes, and subscribers blame the access radios that aren't actually the bottleneck
• Churn from inconsistency: speed tests that look fine at noon and terrible at 8 p.m. produce cancellations and one-star reviews
• Growth ceilings: you can't sell faster plans or add sectors on a tower whose backhaul is already saturated
• Single points of failure: one fiber cut or one failed radio link takes a whole tower's revenue offline until a truck rolls
• Interference-driven decay: unlicensed backhaul that worked at launch degrades as the band fills up, with no contractual recourse because nobody sold you a guarantee
• Misapplied retail circuits: business broadband pressed into backhaul duty hits acceptable-use limits, NAT/static-IP constraints, or simply lacks the SLA a revenue-bearing tower needs

Underneath all of these is a procurement problem more than a physics problem. WISPs often buy backhaul the way they buy everything else — reactively, from whoever answers first, one tower at a time — and end up with a patchwork of circuits, terms, and providers that nobody has a complete picture of. Treating backhaul as a portfolio to be engineered, priced, and contracted deliberately is what separates networks that scale from networks that stall.

The clearest signal is utilization: if your measured busy-hour load on a tower's backhaul is consistently above roughly 60–70% of capacity, you're one growth spurt or one streaming event away from visible congestion. Beyond that, watch for the business symptoms — rising support tickets about slowness that resolve by morning, churn concentrated on specific towers, or a sales team that can't sell your newer, faster plans in certain coverage areas because the infrastructure behind them can't deliver.

Structural triggers matter too. You should be re-evaluating backhaul when you light a new tower or upgrade access radios to higher-capacity gear (modern multi-gigabit access equipment makes an old 300 Mbps backhaul link the bottleneck overnight), when a fiber provider builds into a region you previously could only reach by microwave, when a long-standing unlicensed link starts showing interference, or when a single tower's subscriber count crosses the point where an outage is a five-figure problem.

It's not only WISPs. MSPs and enterprises running private point-to-point wireless between facilities, municipalities and utilities backhauling cameras or SCADA from remote sites, and venue or agricultural operators with campus-scale wireless all face the same question: what carries the traffic once the wireless part ends?

1. Feeding a new tower: a fresh site needs its first middle-mile connection — often licensed microwave from an existing fiber-fed site while a fiber build is quoted
2. Upgrading a saturated tower: replacing an unlicensed PtP link or an undersized broadband circuit with gigabit-class fiber or licensed microwave as subscriber counts climb
3. Adding path diversity: a second backhaul route on a different medium so a single cut or radio failure degrades capacity instead of killing the tower
4. Building an aggregation ring: several towers backhauled to a common fiber-fed hub where upstream internet, routing, and peering live
5. Core and upstream connectivity: the high-capacity circuits (often 10 Gbps or more) from your aggregation point to data centers, transit providers, or peering exchanges
6. Private wireless middle mile: non-WISP operators backhauling campus, industrial, or municipal wireless networks to a central point

Backhaul pricing varies enormously by geography, provider, and path, and anyone quoting firm numbers without a serviceability check and a path study is guessing. What actually drives the number:

• Technology: leased fiber Ethernet is priced monthly per circuit and scales with committed bandwidth; licensed microwave is mostly upfront capital (radios, licensing, engineering, install) plus tower leasing and maintenance; unlicensed PtP is cheap capital with no recurring circuit cost but also no guarantee
• Committed vs. burstable bandwidth: committed-rate circuits bill a fixed tier; burstable services bill against measured usage (commonly 95th percentile), which can punish spiky traffic patterns
• Construction: the distance from the nearest carrier fiber to your tower, and what's in between (rock, water, railroad crossings, permitting jurisdictions), routinely matters more than the monthly rate
• Term length and scale: longer terms and multi-site commitments typically improve pricing; a portfolio of towers quoted together usually beats one-off orders
• SLA level: stronger availability commitments and faster repair response cost more — appropriately, since that's what you're buying
• Tower-side costs: rack space, power, and climb/install labor at both ends of any radio path, plus recurring tower lease fees where you don't own the structure

The honest comparison is total cost per delivered gigabit over the life of the link, including construction amortization and the revenue at risk during outages — not the sticker monthly rate. A microwave link that looks expensive next to a cheap fiber quote can win easily once the fiber's construction cost is amortized, and a 'free' unlicensed link can be the most expensive option on the table once churn from congestion is counted.

A fiber backhaul order typically runs: serviceability check at the tower address (which often means a manual engineering review, since towers aren't in consumer coverage databases) → quote with construction costs itemized → contract → construction and permitting if needed → equipment delivery and turn-up → testing and acceptance. The operator's side of the work is providing site access, rack space, and power; confirming the demarcation point; and having routing ready to accept the handoff.

A licensed microwave deployment runs differently: path study and line-of-sight verification (including a physical path survey where the profile is marginal) → frequency coordination and licensing → equipment selection sized to the availability target → tower work at both ends → alignment, commissioning, and fade-margin testing. The engineering happens before the purchase order, not after — a link designed on paper from elevation data alone can fail in the field over trees that grew since the imagery was taken.

Either way, the deliverable to insist on is a tested, measured handoff: throughput, latency, and loss verified under load before the install is signed off, with the results in writing. That's the baseline you'll troubleshoot against for the life of the circuit.

Timelines vary by provider, region, and path, but the shape is consistent. Where fiber already exists at or near the site, a lit Ethernet service typically turns up in weeks to a couple of months. Where construction is required, expect months — permitting, pole attachments, and boring each add time, and rural builds crossing multiple jurisdictions can run longer. Budget a quarter or more and treat faster promises skeptically until the construction survey is back.

Licensed microwave is usually faster where towers are already standing: frequency coordination and licensing add lead time, but a coordinated path with existing tower access commonly deploys in a matter of weeks to a few months, and it isn't hostage to what's buried under the road. Unlicensed PtP can be live in days — which is precisely why it makes sense as a bridge while the permanent solution is built.

The practical planning rule: start the backhaul conversation when a tower is a drawing, not when it's standing. The tower build takes weeks; the backhaul can take quarters, and a finished tower with no backhaul is an expensive flagpole.

• Sizing to today's peak with no growth headroom — then paying for an emergency upgrade a year later at worse terms
• Buying two circuits for 'redundancy' that share a conduit, a provider, or a single point of aggregation, and learning about it during the cut
• Running revenue-bearing towers on unlicensed backhaul with no migration plan as the band fills up
• Using retail business broadband as backhaul and discovering acceptable-use policies, CGNAT, or missing SLAs at the worst moment
• Accepting a fiber quote's monthly rate without reading the construction line item, the term, or what early termination costs
• Designing microwave paths from desktop elevation data without verifying line of sight and fresnel clearance in the field
• No baseline: skipping documented throughput and latency testing at turn-up, leaving nothing to troubleshoot against later
• Forgetting the tower side: power budgets, rack space, grounding, and climb access solved after the equipment arrives

1. Is this a committed-rate or burstable service, and exactly how is usage measured and billed?
2. What is the actual construction cost to my tower site, itemized — and who owns the risk if the survey finds more work than the desktop estimate assumed?
3. What does the SLA guarantee: availability percentage, repair response time, latency and packet loss targets — and what are the credits when it's missed?
4. Is the physical path diverse from my existing circuits, or does it share conduit, poles, or aggregation facilities?
5. How do I upgrade capacity later — is going from 1 Gbps to 10 Gbps a config change, a new contract, or a new build?
6. What's the term, the renewal mechanics, and the early-termination schedule month by month?
7. Who is the escalation contact at 2 a.m. on a Sunday, and is there a network operations center that will see the outage before I do?
8. For microwave work: who handles frequency coordination and licensing, what availability is the path engineered to, and what fade margin was used?

The choice is rarely one technology forever — it's which technology for which tower, and what the failover path is. Fiber wins on capacity and reliability per dollar where it already reaches; licensed microwave wins on economics and control where it doesn't; unlicensed wins on speed and cost for small or temporary situations; satellite has become a legitimate tertiary backup for sites that genuinely have nothing else.

Wireless ISPs are the obvious case, and within that world the patterns repeat: the rural operator extending from a fiber-fed anchor town outward via microwave hops, the suburban fixed-wireless provider upgrading tower feeds as gigabit access radios come online, the municipal or cooperative broadband project that needs middle mile before it can serve anyone at all.

Beyond WISPs, the same engineering shows up anywhere the last stretch is wireless. Logistics operators backhaul yard and warehouse wireless across sprawling sites to a central connection. Manufacturers connect plants and remote facilities where running fiber across a rail line or a highway is prohibitive. MSPs building private point-to-point links for clients face the identical fiber-vs-microwave decision. Utilities, agriculture, and public-safety networks backhauling sensors, cameras, and remote sites are buying the same middle mile under a different name.

We're a technology advisor, not a carrier — we don't sell you our network, because we don't operate one. For a WISP or tower operator, that means we check serviceability across multiple providers at each of your sites (including the manual engineering reviews tower addresses usually need), quote real pricing with construction costs surfaced instead of buried, and compare the fiber option honestly against what licensed microwave would cost to own. We work with leading technology providers, including carriers and aggregators that specialize in wholesale and middle-mile connectivity, and we manage the order through installation so construction milestones, permits, and turn-up dates don't become your second job.

Because we're paid by the providers, the advice doesn't add a line to your bill — you get an advocate who's seen hundreds of circuit orders and knows where the traps are: the conduit that isn't actually diverse, the SLA credit that isn't worth the paper, the burstable billing clause that punishes your busy hour. Bring us your tower list and your growth plan; we'll bring back options, numbers, and a straight answer about which towers should be fiber, which should be microwave, and which need both.