It's interesting to think about how North Korea could (not) defend itself. The tactical challenges it faces are in part challenges a great power's military would face as well if it's at war with another great power.
Forget about it. A few fishery protection police boats may make sense and might be able to drop some mines in the vicinity of harbours during the prelude to a hot war, and that's about it. All other naval assets are fodder or inconsequential. Effective and survivable coastal defences would be too expensive.
Same. Maybe a few helicopters with unusually skilled or lucky pilots might skim the valleys for a while and avoid destruction, but the fixed wing aircraft are mere targets. North Korea might maintain a squadron of attack aircraft for an uprising scenario and a squadron of interceptors for air policing as well as a few semi-civilian transport aircraft for sensitive international freight.
Area air defence does almost inevitably depend on radars and would thus be subjected to sophisticated anti-air defences campaigns. A few such (mobile!) systems may make sense purely in order to degrade hostile offensive air power potential by forcing some caution on them (including prevention of persistent drone patrols), though.
It's largely a political weapon (threat to Seoul). I suppose the heavy artillery would be consumed almsot in its entirety by the first 48 hours of warfare and the vast majority of other artillery would be gone within a week.
A few heavy (= long range) multiple rocket launcher systems, optimised for camouflage, might remain useful (especially with scatterable AT mines). Their use would depend on the ability to tell when the skies are clear, though. This requires at least local air surveillance (possibly by upwards-looking infrared and UV sensors as well as passive radars).
South Korean air bases and civilian airports would probably be overcrowded and might be enticing targets for bomblet munitions, too (to be released before C-RAM systems can intercept).
I'd focus entirely on defensive tasks; build underwater and pontoon bridges, lay minefields, blow up roads, cause debris avalanches and the like. Forced river crossing and clearing of minefields under fire would be pointless.
Similar to artillery, this would be consumed really quick. Some armour is still invaluable for representing the armour threat during training for the infantry.
The classic North Korean infantry of 1950 was exceedingly impressive at infiltration attacks - even support troops pressed into infantry missions. They were willing to risk much and it paid off.
Infiltration attacks are now exceedingly difficult because the cover of the night was largely lifted by night vision devices.
Still, the infantry is the most elusive force component and needs the least supplies. It would be the most promising (though not the most impressive to civilians) force for strategic defence.
The next part will be exclusively about how to set up a respectable North Korean infantry.
First, ensure morale is adequate. You cannot fight without adequate morale, especially not with duel (line of sight combat) troops. The people of NK need to see improvements in their lives. Every quarter some progress. Some more relectricity, some more or better food, some more clothes, somewhat higher misdeamor requirements for being sent to labour camps. The troops would be very unreliable if they cannot be motivated. Reallocate resources from the useless air force and navy towards consumption and economic investments to create progress.
Second, ensure the NCOs and officers understand Western sensors and the camouflage requirements. Visible / near IR / IR camouflage against ground forces, Visible / IR / radar camouflage against aerial threats. Same for concealment and deception. This is about training, discipline and also equipment.
Third, they need night vision; mostly cheap passive near IR (low light, starlight), but also some passive IR (thermal) sensors for officers and mortar forward observers. Also enough munitions to provide illumination to near IR night vision devices if there's no moonlight.
Fourth, they need to be able to take on main battle tanks. British and American forces have demonstrated how demoralising and irresistible de facto immune MBTs are in ground combat. They depend on them, too. South Korea has comparable MBTs which are a little better optimised for Korean terrain.
Infantry battalions should probably get the Chinese 120 mm battalion recoilless guns for taking on MBTs and for line of sight fire support during local (counter) attacks. The regular infantry squads should be able to make do with the Russian 105 mm warheds for RPG-7 (PG-7VR series), which would be dangerous to IFVs all around and to MBTs from flanks, rear or high ground. The fuzes would need to react to bar (cage) armour in time, keeping the warhead from being deformed prior to ignition.
Fifth, at least reliable VShoRAD (very short range air defence) is a necessity. They would get shot up even by circling Apache helicopters and their 30 mm guns if there's no reliable VShoRAD. Passive infrared is the typical guidance method, but it's insufficient. Countermeasures are numerous and IR-based VSHoRAD ("ManPADS") have lost a lot of respect during the last 10-15 years. Laser beam riders are the way to go, preferably with impact fuze instead of jammable proximity fuzes. The launcher needs to have night vision and support by a passive early warning device. An IFF (identification friend/foe) component is unnecessary.
Sixth, they need mortars dedicated to survival. The hilly and mountaineous Korean landscapes are very difficult terrain for mortar radars, but it still takes skill (education) and possibly some dedicated mortar bombs to manage the risks. The relevant mortar bomb radar cross sections should be kept small, and trajectories need to be kept low. The typical 82 mm mortars and Chinese 'Jet Shot' commando mortars could be used a lot. 120 mm mortars could be used preferably for guided munitions (which make interpolation of origin difficult if they manoeuvre early on already). It may be that mortar teams would still be forced to shoot primarily from within settlements (which is no war crime as long as the civilians are free to leave), as the South Korean government would probably not want many North Korean civilians massacred..
Seventh, they need communications. Motorcycle couriers, improvised cable and fibre-optic land lines, and digital radio relay networks using many dispersed nodes (too many to shoot at with artillery after triangulation) might work. The best approach would probably to simply keep the distances short and the dependence of bandwidth minimised. One-time pads should be used for most radio communications within regular networks (along the ordinary chain of command from company upwards).
Eighth, jamming and decoying should be rampant and be based on many cheap items instead of few expensive ones.
Ninth, bulletproof plate inserts need to be defeated. They defeat AK-pattern assault rifles and that's going to be both unsatisfactory and persistent. Assault rifles still have their place in the infantry, but riflemen with powerful-enough semi-auto long cartridge weapons and quality AP bullets are required, too. This is a must-have for morale, if nothing else (you can also shoot legs, after all).
Tenth, I suppose their army would be most resilient if it was based on battalion battlegroups. A small HQ element, a signaller/courier platoon, a VSHoRAD platoon, sniper platoon, engineer platoon, mortar company, MI/MP platoon, gun company, supply company (no trucks), three or four rifle companies and one replacements/training company.
These battalion battlegroups would receive the definition of their area of operations, orders for their activity there and one-time pads and other information for cooperation with neighbouring battlegroups. At least three different battalion battlegroup types might make sense; militia (settlement defence), regular army (countryside) and strategic army reserve (trained for offensive actions, not the least against paras).
This kind of army forces would not "win" a war, of course. South Korea and Mainland China are both superior to North Korea on their own. They could only make it bloody*, relevant foreigners might be able to anticipate this and this in turn might keep the
peace ceasefire. Or such an army might buy enough time till allies arrive.
I wrote years ago about how military forces seek to minimize the active tactical repertoire of their enemies. The incredibly poor opposition in Iraq and Afghanistan demonstrated failures to maintain the very basics of tactical repertoires. Effective anti-tank defence, effective short range air defence, adequate morale, defeat of NIJ level IV body armour by at least some small arms, uncompromised basic communications, basic night vision and sufficient camouflage are not negotiable. An army hoping to resist sizeable Western-style forces for more than a few weeks needs to possess and maintain these capabilities. It's possible and depends on relatively few expensive hardware items. Reliable VShoRAD missiles are probably the most difficult to come by, for many of the known examples are from Western sources. The Russians use the technology as well, though (SOSNA system, SOSNA-R missile).
related: 2012-08 North Korea's military capabilities and defence (about the impression-focused defence strategy which they actually employ)
*: The regime would probably need to relocate to Hyesan, a Northern border town which is in the mountaineous region that's relatively suitable for defence by light forces.