Especially at the beginning of my career I was much more a system level engineer than a circuit designer, let alone involved with technology. The times of the radar technology (SLAR for remote sensing) and optical communication (coherent systems) were primarily focused on building and/or analysing the complete system. Building electronic units was at that time only a means to come to improved systems. In both cases I have always spent a lot of effort on theoretical analysis and/or performance simulation, where my preference was clearly on the analytical approach. Although this often requires clever simplification, once one obtains a closed and manageable analytical expression for the relevant system behaviour it is possible to extract a lot of information about the system performance from it. My theoretical analysis of coherent receivers started because I wanted (needed) to explain the sensitivity penalties I measured, and subsequently developed into a full analytical approach covering all kind of different modulation and detection techniques. No greater satisfaction than after many days and evenings trying to solve a complex integral with higher-order Bessel functions finally finding a closed expression that could be used directly for calculating our system performance!
The next phase was, mainly in the tuner period, was more involved with defining new RF functionality. Often on the interface between discrete and integrated functions, but later also related to the analogue-digital boundaries: where to digitize. All these discussions were strongly dependent on technology choices and cost-performance analyses.
One nice example was satellite reception, where on the one hand silicon integration became possible (see the TFF1000 in the RF-IC section), while at the same time we were struggling with the increasingly high number of RF cables between the LNB and Satellite Set Top Box (STB). And thirdly digital satellite (MPEG-based) was rapidly replacing analogue satellite. A resulting key patent was the Integrated Outdoor Unit (IOU), with the digitization step moved into the LNB, and a single high speed (Ethernet) digital serial link to the STB. That was 2000, and ten years later these IOU's were on the market!
We had similar discussions (or more heated debates) about the introduction of digital interfaces (MIPI) in the mobile phone cellular front end, with all its consequences for technology choices (RFCMOS required for the transceiver) and system partitioning (no RF-BB one-chip SoC).
My last contribution to the mobile phone market was to trigger the discussion about agile (software-defined) radio's to cope with the explosion of frequency bands and antennas in the phone.
The first phase of system involvement was related to promoting new systems, applications and standards.
The first one was RF Heating, where solid state (LDMOS or GaN) RF power amplifiers are used for microwave heating and cooking, replacing the classical magnetron and even less efficient sources. Although strictly spoken no communication system, it required a new system architecture, control SW and RF control and feedback loops. For a number of years I was the internal NXP champion for RF Energy, assuring budgets and keeping the innovation alive. It is good to see this has now been pulled to the level of an international consortium promoting this application, the RF Energy Alliance.
The second area was the Internet of Things, and especially the drive to get the underlying communication standards harmonized. As NXP we were relying on Zigbee (IEEE 802.15) as low power 2.4GHz packet-based mesh networking standard. To promote and align on the lighting part of the IoT we actively participated in both the Zigbee-Alliance and the The Connected Lighting Alliance, where I was the first NXP board member.
The future will learn if, how and when both RF Energy and the IoT will become a success. One thing will be clear: it will be different than what we thought back in 2012-14.