Bridging the Gap Between Commercial Electronics and DoD Requirements

Being involved in the OUSD (R&E)’s SHIP (State of the Art Heterogeneous Integrated Packaging) program, and internal effort to transition these new electronics to BAE Systems programs, I wanted to write about my observations working with the commercial electronics industry, and the critical steps the Defense Industrial Base needs to take to ensure electronics advancements are relevant for our uses. Hopefully we can start a meaningful discussion around solving some of the challenges the DoD is facing.

Unlike almost every commercial product or system, advanced defense electronic systems range in lifetime quantities of tens to hundreds of thousands. As part of its goal to maintain technical superiority over our adversaries, the defense industry develops bleeding edge technology to accomplish a whole host of missions. For example, the Striker II Helmet Mounted Display may see quantities in the thousands, while the APKWS® laser-guidance kits may see quantities of hundreds of thousands. Other systems, such as the AN/ALR-94 EW Suite, which is critical to mission success for the F22 Raptor, will see quantities in the hundreds over their lifecycle. In contrast, Samsung shipped 58.2 million smartphones in the 4th quarter 2022, while Apple shipped 72.3 million iPhones over the same period. That's ~803,000 per day or ~9 per second. With recurring quantities like this, it's not surprising there is massive investment in commercial SoC (system on a chip), with custom ASIC devices like the Axx series Bionic chip, containing custom CPUs, GPUs, and Neural Engines designed specifically for Apple applications. Bionic is developed (and updated nearly annually) in the world's most advanced semiconductor nodes. Developments of this magnitude are amortized through recurring sales of the end product.

DoD doesn't have access to the Bionic chip, nor would it necessarily benefit from its availability. There are general purpose electronics being driven by embedded computing and data centers that the DoD can and does use in its systems. Examples are devices like the Nvidia Ampere GPU line, the Xilinx Virtex and Versal series FPGAs, and the Intel Agilex FPGAs along with countless others. Multi-chip packages are the way of the future in reducing size, weight, power, and cost of microelectronics by combining multiple chips with ultra-short reach interconnects between disparate semiconductor nodes.  The massive investment made by commercial companies is derived from the massive scale of sales for automotive, consumer electronics, data centers, and cloud computing. The underlying process node investments by commercial foundries occur at a pace and scale the Defense Industrial Base does not and cannot operate at for a multitude of reasons.

The DoD systems leveraging commercial technology make significant performance improvements year over year. However, there is a cat and mouse game at play. As we improve our system performance (be it RADAR, EW, Communications, Computation, etc.) so does the adversary improve their systems that counter or avoid our impact on their mission because this technology is available to anyone in the globalized world we live in. 

There is a solution in this game, which is developing DoD-specific electronic components to technically outmatch adversaries upgrading their systems with commercial electronics alone. In the June 2020 edition of Microwave Journal, I wrote an article (Bridging Commercial and Defense Technology to Maintain EW Innovation) laying out why DoD Electronic Warfare systems cannot rely on commercial investment alone, in part because of the necessity to maintain performance over a wideband frequency range.

In my area of technology, the DoD-specific technology usually takes the form of application specific integrated circuits (ASICs) which have better performance (bandwidth, sensitivity, phase noise, DSP capacity, and RF power) than their commercial counterparts. The DoD sometimes funds these developments through programs of record, but more recently programs like SHIP Digital and STEAMPIPE (OUSD(R&E)) [RP(1] are funding packaging and ASIC development that can be used by the defense industrial base to gain a technical advantage across multiple programs of record using the same technology. The Government makes these surgical investments to keep the edge, but we can't rely on them alone.

DoD investment in ASICs is not new, and they are successfully integrated with commercial electronics in many of today's systems. The interconnection between the ASIC and a commercial chip is through well-defined interface standards that work for both commercial and DoD.

Defense industrial base needs to advocate for itself in these standards bodies as they develop new interfaces. While we won't be the strongest voice in the room due to our relatively low volume needs, strong advocacy in areas of commercial and defense overlap can ensure the Government's surgical investments are cross compatible with the heavy commercial investment. For example, the SHIP Digital program invested in Intel MCPs (multi chip module) and delivered a defense-relevant package to BAE Systems for integration in a pre-EMD program. EMIB (Embedded Multi-die Interconnect Bridge) is Intel's ultra-short reach low power interconnect between chiplets in their MCPs. While it is very high performance, it is not ratified by a standards body such as its cousin JESD 204c, which interconnects devices across a digital circuit board in a low latency, coherent and deterministic transfer.

UCIe is an open specification for chiplet to chiplet interconnects, but the Rev 1.1 standard does not include deterministic data transfer or guaranteed data transport that is necessary for our EW systems. My colleagues proposed migrating the JESD logical protocol over UCIe in a February '23 paper titled "Evolving JESD to UCIe – Ensuring UCIe supports deterministic functions."

Reliance on non-deterministic data transfers will add complexity to systems that require precise alignment of data, such as in interferometric measurements (Direction Finding, ESM), multi-channel radar with multiple beams, and digital phased arrays.

We're looking to identify applications in the commercial or dual-use market that require the same deterministic data transfers, to influence the UCIe specification so it is more aligned with the needs of DoD systems. Whether you are working for one of BAE Systems competitors, or you are in the commercial industry with a concern about National security of the Defense Industrial Base supply chain and technical edge, your thoughts and actions can help us achieve this goal as a group. Neither the Government nor any individual company can do this alone, so we need your help.