Satellite & Hybrid Connectivity
Engineering satellite into ISP networks the right way — as policy-controlled backhaul, redundancy, and rural reach, not as an afterthought bolted to the network edge.
Satellite belongs in the design, not on top of it.
- VSAT
- GEO HTS
- LEO
- Multi-Orbit
- Ku/Ka-Band
- Teleport Operations
- RF Engineering
- BGP
- Policy-Based Routing
02 — Where It Belongs
When Satellite Belongs in the Design
Satellite earns its place in an ISP network in three roles: backhaul where fibre and microwave do not reach or do not pay back, redundancy that shares no failure domain with the terrestrial path, and rural reach where the economics of trenching or towers simply do not close. The common 2026 mistake is the opposite framing — treating LEO as a competitor to defend against, rather than a transport class to engineer with.
A satellite link is not a slower fibre link. It has its own latency profile, its own weather behaviour, its own contention model, and its own failure modes. A network that ignores those differences converts a resilience investment into a customer-complaint generator.
Our work is putting satellite where it belongs — and integrating it so that subscribers cannot tell when it is carrying their traffic.
03 — Domains
Satellite Engineering Domains
From the routing policy down to the feed horn
Hybrid Integration & Traffic Policy
- Routing policy per application class across satellite and terrestrial paths
- BGP path control, conditional advertisement, and asymmetry management
- Satellite-specific performance baselines (latency, jitter, loss under fade)
- TCP behaviour over long-delay paths — window scaling, PEP placement, and their limits with encrypted transport
- SD-WAN overlays that respect the underlying link character
VSAT Deployment & Commissioning
- Site survey, look-angle, and obstruction analysis
- Antenna pointing, polarisation alignment, and cross-pol isolation
- Carrier line-up discipline with the satellite operator
- Commissioning checklists and acceptance criteria per site
- Installer capability building and quality control at scale
Teleport & Hub Operations
- Hub infrastructure design and outbound/inbound carrier planning
- Bandwidth management and QoS enforcement at the hub
- Terminal fleet monitoring and proactive fault detection
- Capacity and contention management across the terminal population
- Coordination between hub, NOC, and field teams
RF & Ground-Segment Engineering
- Link budgets with realistic rain-fade margins for Ku and Ka band
- Cross-pol interference (CPI) diagnosis and normalisation
- Interference identification and escalation to the operator
- Large-aperture earth station recovery and satellite tracking configuration
- Ground monitoring of satellite fleets and carrier health
LEO / GEO Blending Per Site
- Orbit selection by application mix — LEO for latency-sensitive classes, GEO for bulk and broadcast
- Multi-orbit failover ordering and pre-staged policy
- CGNAT and egress-path implications of LEO services
- Cost-per-site modelling: terminal, service, and operational overhead
- Avoiding single-vendor dependency in the space segment
Failover That Does Not Degrade Service
- Failover policy decided in advance — which application classes ride the satellite, which shed
- Capacity triage so the backup link is not crushed at switchover
- Session-survival engineering across path changes
- Scheduled failover testing, not faith-based redundancy
- Baseline-driven alarms that distinguish fade from fault
04 — Integration
Policy-Controlled Integration
Satellite fails ISPs when it is treated as generic bandwidth. It works when the network is told, explicitly, what each link is for.
Routing Policy per Application Class
Voice, interactive traffic, and control-plane sessions are steered to the path whose latency profile they tolerate; bulk transfer, caching fills, and backup traffic take the capacity-rich path. The policy lives in the routers, not in a hope that congestion sorts itself out.
BGP Path Control
Deliberate path selection and advertisement across satellite and terrestrial upstreams — so that failover is a routing decision the network has already rehearsed, and return traffic follows a path the design actually anticipated.
Satellite-Specific Baselines
A GEO path at 500+ ms round trip is not broken, and a Ku-band link in a rain cell is not down. Monitoring baselined per link class prevents both false alarms and the quiet degradation nobody notices until customers do.
This is not a whiteboard pattern. We ran it from inside a satellite upstream provider — integrating satellite into a hybrid ISP core through policy-based routing by application class, BGP tuning, and satellite-specific performance baselines, with RF coordination across Nairobi, London, and Washington hubs.
Satellite does not degrade networks. Pretending it is fibre does.
06 — Track Record
Where This Experience Comes From
Every recommendation on this page traces to field work — teleports, hub infrastructure, and end-user VSATs:
- Six years at Belgium Satellite Services delivering senior technical support to ISPs across West, East, and Central Africa
- Satellite and infrastructure overhaul for a Kenyan satellite broadcaster, working alongside SES Newskies, Vertex, and TST engineers
- Recovery of a 7.3m earth station for UNSOA in Mogadishu — normalising unacceptable cross-pol interference and configuring satellite tracking
- Ground monitoring of the Avanti satellite fleet from stations in Kenya and Tanzania
- BBC World Service distribution logistics across 73 partner stations in Nigeria and Somalia
- The UN IOM PIRS border-control network for Djibouti — satellite, optics, and microwave in a single hybrid design
- Healthcare connectivity for Jembi Health and the Rwanda Ministry of Health; deployments for CEPSA (oil) and KenGen (power)
- Satellite upstream operations at InterSAT — core network and server administration with RF coordination across Nairobi, London, and Washington hubs
- Formal Avanti VSAT installation and commissioning training (2013)
We have operated all three sides of the link — the teleport, the hub, and the remote terminal.
Multi-orbit is now the operating assumption, not a novelty: Starlink spans some 26 African markets (with three countries holding roughly two-thirds of subscribers) while Eutelsat/OneWeb and SES sell blended GEO+LEO capacity, so the per-site question has shifted from “satellite or not” to “which orbit for which application class”. The catch is that the old satellite toolkit only half-transfers — the TCP performance-enhancing proxies that rescued GEO paths cannot see inside encrypted QUIC, and LEO handovers punish congestion-control algorithms tuned for stable long-delay links — which is why per-application routing policy and satellite-specific baselines matter more, not less, than in the pure-GEO era.
Bring Satellite Into the Design Properly
Start with a focused review of where satellite sits in your network today — and where it should.