Science gives you theories and discoveries, but engineering is what makes them useful. Amr Ibrahim draws that distinction almost in passing, near the end of an hour-long conversation, and it sounds almost like a personal philosophy. The Nokia system expert has spent thirteen years in telecoms on the engineering side of that line, making sure that systems work. On land, at sea, and in the hands of humans.
Private Networks (4G/5G) System Expert at Nokia
Amr Ibrahim's answer to what makes a network reliable is slightly inconvenient for anyone hoping for a single fix: it depends on the interplay of all its components.
Take this example. Two antennas, one that listens and one that transmits, were mounted too close together on a job at an oil and gas site in the middle of nowhere. Everything should have worked. The signal was there, the configuration checked out, the software gave no errors. But it didn't work . Amr's team drove three to four hours to find out why, and when they arrived, the fault turned out to be almost physical. The two antennas were close enough that the transmitting one was picking up its own signal, feeding it back into itself, amplifying noise instead of coverage. "And that tells us that whoever did this didn't know the system," Amr says.
Amr Ibrahim is a Private Networks (4G/5G) System Expert at Nokia in Finland. He got to that antenna story by way of CommScope, where he designed distributed antenna systems, DAS for short, the passive infrastructure that carries an outdoor signal into places like buildings, airports, and stadiums where it otherwise couldn't reach. Passive, because the signal travels only through coaxial cable and splitters, no amplification, no active electronics, just careful routing designed to lose as little power as possible on the way in.
At Nokia, his job moved further down the stack, into the hardware that generates the signal itself: the radio units, and the many parameters that need setting correctly before a system ever reaches a customer. Where CommScope was about the best way to distribute a signal, the goal at Nokia is about configuring, testing, and troubleshooting the thing that creates it in the first place.
Ask Amr where a private network is most likely to fail and he won't point at software first. He'll point at the simple things. A seal. A connector. What he calls a thermal path, the route heat needs to take to get out of active equipment before it does damage. "If you're choosing the wrong material with the wrong thermal requirement," he says, "that's a critical thing. We're talking about active equipment, so heating is critical, and that's something many people aren't paying enough attention to."
He has seen customers spend millions on hardware and then hire the cheapest available contractor for the install. "You're saving maybe tens of thousands on a million-dollar project. That doesn't make sense," he says. "They don't understand what the cost of this is." A radio found full of water on site. A splitter installed the wrong way round because two ports that looked identical weren't. Two antennas facing each other where they should have been kept apart. None of these are software problems. All of them take the network down anyway. As Amr puts it, reliability depends on "the power, the software, the installation, the cell load, what kind of applications will run on the server considered together."
Think of a plumber. Twenty years into fixing pipes, a good plumber can often diagnose a kitchen's most likely fault in thirty seconds, because the unglamorous installations have a tendency to fail in the same handful of ways, again and again. Amr's version of that instinct is built on years of seeing the same faults repeat themselves in different forms: cables, seals, connectors, and cooling systems. "No material fixes a system installed by someone who doesn't understand it, and no amount of understanding can replace the right material for the job."
Amr's other, less expected credential is a Tampere University Master's thesis where he used machine learning to detect high latency incidents in private networks. The site was a smart port, where autonomous shuttle carriers move containers between ship and shore under a safety protocol that halts them the moment latency crosses a threshold. Even a few minutes of a stalled shuttle carrier costs a port real money.
The incidents worth catching turned out to be vanishingly rare, buried inside the ordinary pattern of network behaviour. "Machine learning can't do magic," Amr says. "The problem is that it's easy to get into Python, data science, data analysis, machine learning, and algorithms. But it's hard to acquire the domain experience that is needed to make actual sense of the data patterns. If you don't understand the domain you're applying it to, you'll most probably get the wrong conclusions."
His favourite illustration of why is a well-known one in data science circles. "There's a machine learning algorithm which found that in some places, the percentage of people drowning is proportional to ice cream sales," he says. "If you sell more ice cream, more people will drown." Taken at face value, it's nonsense, and everyone knows it's nonsense, because everyone already understands the real cause: "Most probably people are swimming in summer and eating ice cream in summer," he says. "But correlation doesn't mean causation."
That, to Amr, is the real story of AI in his industry. Not that machines are replacing engineers, but the fact they're making domain expertise more valuable than before. Machine learning, in his words, is not wizardry. It surfaces patterns. Whether those patterns mean anything still depends on someone who has spent enough time in the field to know the difference.
Ask Amr what the next five years of private networks look like and the answer is about who's in the room on the customer side. Right now, he sees a persistent gap, because a reliable operation will take more than classic IT department skills. "BTS, radio units, 4G and 5G, are different from routers, switches, and IP networking. That's part of the telecom network, but it's not the telecom network. Without someone in-house who understands both the operational environment and the technology being deployed, customers can end up investing in systems they don't fully understand the value of."
The customers who get the most out of their private networks, in Amr's experience, aren't necessarily the ones who spend the most.
Thirteen years in telecoms and a Master's thesis on the side is a lot of investment in one field. "I'm just curious about learning new things," Amr says. "That's why I'm doing an MBA now, even though it's a bit far from my current role. I want to explore things outside the engineering world, like how business is run." He puts that curiosity down to an engineer's mindset.
"We engineers have that nature," he says. "We are all problem solvers. We're drawn to challenges."
Amr Shamel Ibrahim is a Private Networks (4G/5G) System Expert at Nokia in Finland, with more than 13 years of experience in the telecommunications industry. He previously held positions at CommScope before joining Nokia in the network systems area, giving him working experience of both the passive infrastructure and the active systems sides of the network. His work sees him delivering and commissioning private mobile networks for enterprise customers, where reliability, latency, and the physical dependencies underneath the software layer all matter directly to whether the customer's operations run.
Amr Shamel Ibrahim
Private Networks (4G/5G) System Expert at Nokia
Explore the edge
Mind-boggling and horizon-broadening, Uniquely Wired is your access to interviews with thought leaders from the brands that define tomorrow. Pick their brains now.
- Articles
Do you want to feature in the next edition of Uniquely Wired? Our editors are always looking for new stories. Is yours exciting?