New Science Project - WiFi/BT scanner and Security Tool

[sfx2000]

Inspired by the article here.... floating out an idea...

As part of the poll - some options are below the price to develop, some break even, some give me a bit of profit....

I can do it - handheld and battery powered, with a local display, and a couple of buttons to change modes, totally configurable on the WiFi stack for 2.4GHz... anything from passive scanning to active changes on the network. Also can cover Bluetooth - both regular 4.2 and Bluetooth LE. Might include an ethernet interface, but it will definitely include a USB - both as client and host via USB-OTG

Current form factor plan is similar to the old-school Nokia 6190 or smaller - the board I'm planning is about half the size...

Software will be open, I'll publish this on Github - it's not linux, but if one knows Arduino, it'll be familiar enough.

I've already worked out the BOM and NRE going in - just curious as to the market interest before I move forward.

Some history - I'm an experienced project engineer and product manager for a major handset OEM, and built and maintained software platforms for one of the top Wireless Carriers in the US.

Side Projects - I've successfully built other devices, both HW and SW - last project I did was successful as a business exit.

 

[thiggins]

All depends on the feature set. What will set this apart from app-based scanners?
The key feature I would pay for is channel activity/congestion.

Did I read correctly that it will cover 2.4 GHz only?

 

[sfx2000]

All depends on the feature set. What will set this apart from app-based scanners?
The key feature I would pay for is channel activity/congestion.

Did I read correctly that it will cover 2.4 GHz only?

For now because of the radio in the module* - it can do WiFi b/g/n and Bluetooth in the 2.4GHz band.

* module as it's much easier to pass FCC testing as the module is already approved - antenna is contained within the module.

If I go dual-band - it's a lot more work for regulatory approvals, and the cost would go up - war gamed things out, and we get out of scope of what the original project was intended, and the cost goes up significantly - well above $100USD and in a limited run, let's say 1K units, well above that.​

Anyways - tinkering about with Kicad, and the board layout - think the size of a laptop drive... 100mm * 70mm * 12mm roughly - gives a lot of room for the battery and screen - that includes some button actions (on/off, dkey, action, back buttons) - I'm talking with a couple of old-school cellphone CAD guys there - basically make as much compression fit at possible - but this complicates things if I go with plan "B" (DIY kit - see below)

The SW is a lot easier - the module vendor has libs already have some support what I'm targeting, along with the display drivers. I've got the module code up and running now, and it can discern neighbor SSID's, the clients it detects that are associated with those SSID, and make some inference on activity... ideally, it'll do a scan and make recommendations on which channels are best to support.

FWIW - the firmware has two MAC addresses - one for Client STA, and one for SoftAP, which supports Ad_Hoc IBSS - which can be used for firmware updates, software packages, and attaching to a phone/laptop, it can be the remote interface​

Anyways - from a biz perspective, which always comes into product definition - at a limited run of 1,000 units, 50 bucks is doable to cover the NRE and production costs, with a bit to spare and break even. The big cost driver is actually the tooling for the housing...

I'm looking at options of doing it without the housing as the DIY kit, e.g. the module is mounted on the mainboard, along with the appropriate connectors, and the display/battery is in the box... and this brings down cost quite a bit.

I'm exploring other options - Pi HAT perhaps, but there, that's already been done, unless I add some additional functionality like RTC/GPS, which many haven't done yet.

 

[thiggins]

I'd say you need to be able to do at least what MetaGeek did with their $99 Wi-Spy. Better if you really can judge how busy a channel is.

Definitely do it as a DIY kit.

Still not sure of the appeal of a 2.4 GHz only product.

 

[sfx2000]

I'd say you need to be able to do at least what MetaGeek did with their $99 Wi-Spy. Better if you really can judge how busy a channel is.

Definitely do it as a DIY kit.

Still not sure of the appeal of a 2.4 GHz only product.

The module I had in mind doesn't do the WiSpy kind of detail - the 8266 is pretty cool, and has a lot of community support and a good feature set.

WiSpy kind of functions - The ChipCon 2500 from TI can... it can do the raw transceiver stuff, and report RSSI across a selectable bin width... 200KHz is the default there, which would be sufficient - if I recall, this chip is actually related to the chip in the First Gen WiSpy dongle.

Challenge here is that it needs an MCU (like ez430 which it's commonly partnered with) or a CPU (via UART-USB bridge to a PC, or SPI to a Pi-type of SBC over the SPI bus) - and it's frequency limited, which kind of puts us back into the 2.4GHz only bucket.

Had coffee this morning with an old friend from the Qualcomm days - bounced this project idea off him - good conversation about things - he's also a product development/systems design guy like me, and bounced back and forth...

• Form Factor - Gameboy form factor seems better than landscape mode...
• Pi HAT - seems reasonable... can support the LiPo charging and monitoring circuits, and use the Pi for MMI and everything else - challenge is Pi is a bit power hungry
  • Just do it on the Pi - Pi3B+ is dual band, drivers don't support Monitor mode, but the chipset does, so hack the drivers (note - that's been done already for Android, so just need to port the changes to the driver included with most of the Raspberry Pi devices)
• DIY vs. all delivered - all agree would be fun as a kit - just add the Pi and the HAT, and 3d printer files and code hosted on github
  • So would be a fun project to do, just to keep busy, and touch up on embedded systems stuff, however...
• Limited Utility, Narrow market - so low volume, probably won't break even
• Using an old Android phone and an App checks all the boxes for the original user story

So... while it was fun to war-game it out - not sure I'm going to proceed...

 

[sfx2000]

mocha still a sidebar project, running on the back burner...

Scope has changed a bit - but still a useful thing... looking for a reason to push this into production

Basic arch - ESP32-WROVER-B - it's a system in module - so certs and integration is much easier...

RPI host, and a common enough bus with other SBC's - mostly for communications and programming of the ESP32

Software - the ESP32 does support multiple IDE's - NodeMCU, Arduino, and you can go to metal on FreeRTOS and a fair amount of free/opensource SW there

So for prod - basic schematics are done, board layout with the SOM is done, and found a vendor that can do PCB's at $5/sq in for 2 layer, $10/sq in for 4 layer for proto/P1 runs... so for proto, it's good (one has to consider the BOM as well - so the $25 target is good with Cost of Goods so far)

Anyways - target of what the final board could look like - this is a borrowed representation of what the mocha board might look like...

(note - that's a wish - reality is that the board will not have the ESP32 chip discrete, but as part of a SOM on the board)

 

[sfx2000]

So... while it was fun to war-game it out - not sure I'm going to proceed...

What holds me back on this project - to be successful as a business, it would have to be around 650 units sold, which is a bit of a stretch considering a narrow market... and that would pay for a few months of living expenses...

And the risk here is that once I launch the board - wouldn't be much of a challenge for someone in Shenzhen to clone it and undercut pricing...

Gah - the product doesn't exist, but as soon as it does...

 

[sfx2000]

Quick update - I have a proof of concept up and running, draft schematics, and a prelim circuit board...

Proof of Concept - ESP32 on a breadboard - it works, power, connectivity, storage - all good...
Schematics - I've gone thru this more than a couple of times, so waiting on a peer review from one of my old-school EE colleagues...
CCA - ended up being more than the draft above - still was looking at things there...

But I've decided to stop the hardware side of the project - after looking at the next steps of getting a couple of proto boards done, along with the bill of materials going in - really narrow market...

What stopped it for me was finding the m5stack - took some digging, but it really looks like a better project than mine...

$25USD FOB Shenzen - I can't compete on this for pricing or HW features here...

https://m5stack.com/

 

[sfx2000]

So as you can see - for me - hardware is likely now a non-starter... happy I found it... it's not the only hacker easy device, there are others, but this was the most comprehensive one...

I've been tinkering around with the ESP32 software - and it's kinda fun in an old-school way - nothing ready for production, but the ESP32 has a fair amount of latitude with how it works with WiFi and BT/BTLE, and that should be flashable on the m5 stack...

 

[sfx2000]

Quick update - I have a proof of concept up and running, draft schematics, and a prelim circuit board...

Proof of Concept - ESP32 on a breadboard - it works, power, connectivity, storage - all good...
Schematics - I've gone thru this more than a couple of times, so waiting on a peer review from one of my old-school EE colleagues...
CCA - ended up being more than the draft above - still was looking at things there...

But I've decided to stop the hardware side of the project - after looking at the next steps of getting a couple of proto boards done, along with the bill of materials going in - really narrow market...

If folks want to follow thru on the Pi aspect - here's a decent link...

https://github.com/al177/esp_hat

Still mulling on something moving forward - the ESP's are interesting in many ways - they're useful for IoT applications with a rich set of API's and 3rd party SW...