Your reviewer is very likely an expert in the exact area you are proposing to work in. That is how panels are assembled. So the background section is not a formality you write to fill a page — it is the section where the panel decides whether you know the field well enough to be trusted with the work. An omission you did not notice is an omission they will.
What the section has to establish
Four things, in roughly this order. What the current best approaches are, named specifically and cited. How each of them works, briefly and accurately. Where each one fails against the requirement you care about, in measurable terms. And what remains unsolved — which is the gap your innovation fills.
Weak: "Existing methods are slow, expensive, and unsuitable for field use." A reviewer who spent a decade developing one of those methods reads that as dismissiveness. Strong: "Capacitive probes achieve the required resolution but need a controlled dielectric environment; the two reported field-hardened variants trade roughly an order of magnitude of resolution for that ruggedness, which leaves the field-portable case unaddressed." Specific, fair to the prior work, and it defines the opening you are proposing to enter.
Be generous about prior work. Reviewers are far more suspicious of a founder who claims nobody has tried this than of one who explains why previous attempts stopped where they did. "No one has done this" is almost always wrong, and it is exactly the claim an expert can disprove from memory.
Literature and patents, both
Academic citations alone are a partial picture. A great deal of engineering state of the art is disclosed only in patents and applications, and a reviewer from industry will notice if you have not looked. Cite real, verifiable references: authors, publication, and patent numbers you have actually read. Never cite something you have not opened, and never pad the list — a short accurate reference list beats a long one containing a paper that does not say what you claimed.
The mechanics of searching are covered elsewhere on this site rather than here: start with how to run a prior art search, use CPC classification searching to find the families keyword searches miss, and the free search tools guide covers the databases. Reading a granted patent's claims tells you what is genuinely fenced off, which is different from what the abstract implies.
Commercial alternatives count too
The state of the art includes what a customer can buy today, not only what has been published. A product on the market that solves 80 percent of the problem at a tenth of your projected cost is a serious objection, and the panel will raise it whether or not you do. Address it directly, using the same discipline as a competitor analysis.
Done properly, this section quietly earns points on more than one criterion, since command of the field feeds directly into whether the panel believes your approach, as described in how panels score proposals.
Projects House runs prior-art and technology landscape work as part of early product development. Ask about it through the contact form.