What's the most innovative solution you've come up with for a tricky engineering problem?

During the early growth phase of the mobile industry, I served as Senior Engineer in the Project and Support Department of a major carrier. At that time, prepaid SIM cards were just being introduced and were rapidly becoming one of the key revenue streams.

The market offered two main solution paths:

• Vendor-led CAMEL (Customized Applications for Mobile network Enhanced Logic) platforms — robust but expensive.

• System Integrator (SI) solutions — often based on private PBXs such as Summa Four, providing a more cost-effective alternative.

Since the price gap between Vendor and SI solutions was 4–10 times, I supported the company in managing technical and financial risk by adopting an SI-based prepaid solution. This choice enabled the carrier to enter the prepaid market quickly while balancing cost control and scalability.

However, the SI-based solution came with two major shortcomings:

1. Scalability constraints – As traffic increased, the PBX platform had to expand significantly, to the point where forecasts showed it might eventually scale up to the same size as the mobile core switch, which was neither cost-effective nor practical.

2. System stability issues – The PBX control host became less stable under high traffic loads, since the prepaid application was an in-house, tailor-made development by the SI rather than a mature, carrier-grade product.

To further address these issues, I performed a detailed analysis of both scalability and stability challenges. I broke down the prepaid system into functional layers — Application Program, Database, Circuit Switching, Administration, and O&M.

• From the SI perspective, their main focus was on code reviews for the application layer and fine-tuning the database to improve system stability.

• From my perspective as Ericsson System Support for the carrier, I believed the real bottleneck lay in circuit switching.

PBX was consuming significant processing power to perform circuit switching — mapping each incoming E1 timeslot from the mobile network to an outgoing E1 timeslot back to the network — and maintaining/monitoring those connections.

That led me to ask: What if we could bypass the PBX for switching entirely?

I proposed a simplified approach:

• Directly connect the two E1s at the mobile network side so that when a call arrived on timeslot 1 of the first E1, it would always exit via timeslot 1 of the second E1.

• The PBX would then only need to handle the C7 signalling channel, not the bearer traffic itself.

This significantly reduced PBX load, stabilized performance under higher traffic, and provided a more scalable architecture without requiring massive PBX expansion.

Outcome of the Solution

By offloading circuit switching away from the PBX and letting the mobile network E1s directly connect, my solution effectively eliminated the scalability bottleneck.

• Scalability fixed – Since the PBX no longer had to perform bearer switching, its processing load dropped drastically, making the prepaid system capable of handling growth without requiring PBX expansion to the size of a mobile core switch.

• Simplified architecture – The prepaid application was greatly simplified. No more circuit-switching setup, maintenance, or monitoring was needed, which in turn reduced operational complexity.

• Improved stability – The PBX control host became far more reliable, as it only needed to manage signalling (C7), reducing downtime and minimizing service disruption for prepaid customers.

• Business impact – The carrier could scale prepaid services with confidence, at significantly lower cost compared to a vendor CAMEL solution, while maintaining service quality.

During the early growth phase of the mobile industry, I served as Senior Engineer in the Project and Support Department of a major carrier. At that time, prepaid SIM cards were just being introduced and were rapidly becoming one of the key revenue streams.

The market offered two main solution paths:

• Vendor-led CAMEL (Customized Applications for Mobile network Enhanced Logic) platforms — robust but expensive.

• System Integrator (SI) solutions — often based on private PBXs such as Summa Four, providing a more cost-effective alternative.

Since the price gap between Vendor and SI solutions was 4–10 times, I supported the company in managing technical and financial risk by adopting an SI-based prepaid solution. This choice enabled the carrier to enter the prepaid market quickly while balancing cost control and scalability.

However, the SI-based solution came with two major shortcomings:

1. Scalability constraints – As traffic increased, the PBX platform had to expand significantly, to the point where forecasts showed it might eventually scale up to the same size as the mobile core switch, which was neither cost-effective nor practical.

2. System stability issues – The PBX control host became less stable under high traffic loads, since the prepaid application was an in-house, tailor-made development by the SI rather than a mature, carrier-grade product.

To further address these issues, I performed a detailed analysis of both scalability and stability challenges. I broke down the prepaid system into functional layers — Application Program, Database, Circuit Switching, Administration, and O&M.

• From the SI perspective, their main focus was on code reviews for the application layer and fine-tuning the database to improve system stability.

• From my perspective as Ericsson System Support for the carrier, I believed the real bottleneck lay in circuit switching.

PBX was consuming significant processing power to perform circuit switching — mapping each incoming E1 timeslot from the mobile network to an outgoing E1 timeslot back to the network — and maintaining/monitoring those connections.

That led me to ask: What if we could bypass the PBX for switching entirely?

I proposed a simplified approach:

• Directly connect the two E1s at the mobile network side so that when a call arrived on timeslot 1 of the first E1, it would always exit via timeslot 1 of the second E1.

• The PBX would then only need to handle the C7 signalling channel, not the bearer traffic itself.

This significantly reduced PBX load, stabilized performance under higher traffic, and provided a more scalable architecture without requiring massive PBX expansion.

Outcome of the Solution

By offloading circuit switching away from the PBX and letting the mobile network E1s directly connect, my solution effectively eliminated the scalability bottleneck.

• Scalability fixed – Since the PBX no longer had to perform bearer switching, its processing load dropped drastically, making the prepaid system capable of handling growth without requiring PBX expansion to the size of a mobile core switch.

• Simplified architecture – The prepaid application was greatly simplified. No more circuit-switching setup, maintenance, or monitoring was needed, which in turn reduced operational complexity.

• Improved stability – The PBX control host became far more reliable, as it only needed to manage signalling (C7), reducing downtime and minimizing service disruption for prepaid customers.

• Business impact – The carrier could scale prepaid services with confidence, at significantly lower cost compared to a vendor CAMEL solution, while maintaining service quality.