Nigeria Cameroon Submarine Cable System (NCSCS)

The Nigeria Cameroon Submarine Cable System (NCSCS) is a short, single-owner cable that reveals a textbook story about African internet economics: a 1,100-kilometre fibre linking two neighbours whose actual latency, as measured from the outside, depends far more on commercial routing than on physics. Our 93 measurements over 30 days show a cable that works perfectly in one direction and chaotically in the other — a gap that cannot be explained by the fibre itself.

A 1,100-km point-to-point cable

NCSCS is unusual for two reasons. First, it is owned outright by a single carrier — Cameroon Telecommunications (Camtel), the country's partly state-owned incumbent. Most African cables of comparable length and era are consortium-built. Second, it is point-to-point: only two landings, both in countries that already have more cable capacity through larger consortium systems. NCSCS exists specifically to provide Camtel — and through Camtel, Cameroon — a direct, Cameroonian-controlled path to Lagos, West Africa's largest internet exchange point, without relying on the peering terms offered by consortium cables.

On paper that is a sound commercial strategy. A national carrier that owns its own international cable is not beholden to the pricing decisions of its neighbours. In practice, as our data shows, ownership of a cable does not guarantee that the cable is actually used for every packet.

The one-way cable

We measure NCSCS in both directions between its two landings. The results are striking:

These are measurements of the same two endpoints, going in opposite directions over the same submarine fibre. If the cable were symmetrically used, we would expect both rows to look alike — minimums near the physics floor of 10.8 ms (the cable is 1,100 km long, so pure speed of light in glass gives ~11 ms round-trip minimum), averages close to the minimums, and small standard deviations.

The Lagos → Kribi row is exactly that: 28 clean samples, minimum 19.9 ms, average 20.3 ms, ratio 1.00 against its own stable baseline. The packet takes the cable, arrives, comes back, and our measurement is done. The actual 19.9 ms minimum — roughly double the physics floor — is normal for any real submarine cable: the fibre is not a straight line, it has repeaters every 80 km adding processing delay, and the terrestrial tails at each landing station add their own hops.

The Kribi → Lagos row tells a different story. The minimum is also 20.0 ms, which means the cable can carry traffic quickly in this direction too. But the average is 242.9 ms and the maximum is 383 ms — an order of magnitude more than the minimum. The standard pattern, when a cable has a stable minimum but a chaotic average, is that some packets take the expected fibre path and some take a completely different one. For a 20-ms minimum to co-exist with a 240-ms average, the alternate path has to be roughly ten to twenty thousand additional kilometres of fibre — the length of a return trip via Europe. Our hypothesis is that Camtel's outbound routing from Kribi periodically hands packets to an upstream transit provider (SFR-style intra-AS) that carries them through Marseille or Paris and back down through another cable, instead of pushing them onto NCSCS directly.

Why would a carrier do that if it owns the cable? Commercial reasons. Upstream transit contracts include committed capacity and route preferences that can win out over physically-optimal paths. A carrier like Camtel balances direct costs (paying the cable's own OPEX) against contractual commitments (committed transit volumes on French or UK routes that have to be filled to avoid penalty). When the latter wins, the direct cable is bypassed.

Our probes

We also measure Lagos from outside Africa, which confirms the local story:

Traffic from our distributed probes into Lagos shows stable latencies entirely consistent with European peering: every packet goes to Europe, crosses into a West Africa cable (most likely 2Africa, MainOne, or Equiano) and lands at Lagos. None of them reach Cameroon first. This is the dominant pattern: for most of the world, West Africa is reached through Europe. NCSCS exists to provide a localised exception for traffic that is already in the region — Cameroon to Nigeria specifically.

Why this matters

A 1,100-kilometre cable between two neighbouring West African capitals ought to be a model of efficiency. Instead it is a model of how much of African internet traffic still routes through Europe, even for destinations just across a border. The NCSCS asymmetry we measure — stable 20 ms one way, chaotic 240 ms the other — is a direct fingerprint of that pattern. When traffic originates in Cameroon, Camtel has routing options that do not involve NCSCS at all, and those options sometimes win on commercial grounds. When traffic originates in Nigeria (on networks that are not Camtel), routing back to Kribi over NCSCS is the obvious choice — hence the clean 20 ms.

This is not a failure of the cable or of Camtel. It is a rational outcome of an asymmetric relationship: Nigeria is the regional hub with rich connectivity; Cameroon is a smaller market whose national carrier has taken on the cost of an international cable to preserve autonomy. The cable works physically. Whether it is used depends on commercial rather than physical forces — and it is exactly that sort of insight that submarine cable monitoring, measured from many directions over many days, makes visible.

Try it yourself

Live latency data on the NCSCS cable page. For regional context, see 2Africa (the 45,000-km hyperscaler-led cable that lands in Lagos) and Maroc Telecom West Africa (another single-owner African cable with a similar strategic logic). Our measurements refresh every two hours.

Short as it is, NCSCS remains a meaningful independent link.