Bluetooth presence (via HC05 and Jeenode)

Hacked up a couple of sketches.   BlueTerm which is just a terminal emulator via the Arduino so you can type AT command manually.  BlueNode which is my version of a proximity detector node, I've not done any of the backend LUA sketch work.

http://livebox-hah.googlecode.com/svn/trunk/userapps/arduino/Bluetooth/

Interesting in that Garys sketch only has room for a 10 chars for the Bluetooth device however I found that 12 chars where needed.

Some sample debug output from my sketch.

ASYNC> AT+INQ
+INQ:12:37:E90912,520204,7FFF
device=12:37:E90912 type=520204 rssi=7FFF
+INQ:1D:FE:720479,7A020C,7FFF
device=1D:FE:720479 type=7A020C rssi=7FFF

Responding to myself:

A bluetooth address breaks down like this: NAP:UAP:LAP

From here http://homepages.inf.ed.ac.uk/group/sli_archive/slip0304_b/resources/com...

Addresses

Each Bluetooth device is assigned a 48-bit address consisting of three parts: a 24-bit lower address part (LAP), an 8-bit upper address part (UAP), and a 16-bit so-called non-significant address part (NAP). LAP and UAP address parts are used in several places in the Baseband protocol. The 32-bit concatenation of the upper and lower address parts is also used and is abbreviated as ULAP.

This makes it easier the 32bit concatenation (ULAP) is now the same size a MAC address (4 bytes) and this is all that is required for unique identification.   A mac is easier to push about and use as its familar.....

Time to sleep - I will rework this payload with this in mind, that will simplify the backend sketch too.

Gary - I'm getting up to speed myself and I did examine your code for some guidance so thanks for the trail blazing it saved me some research time.

I updated my .INO sketch so the payload represents the device format more succintly.  Sending a variable length ASCII string just doesn't feel right to me.

// RF Transmission payload.
struct {
  union {
    struct {
      unsigned int nap:16;
      unsigned int uap:8;
      unsigned int lap:24;
    };
    uint8_t device[6];
  };
} payload;

We can break apart this device string into the payload struct for transmission.
A bluetooth device address breaks down like this: NAP:UAP:LAP

+INQ:1D:FE:720479,7A020C,7FFF

      uint16_t nap;
      uint8_t uap;
      uint32_t lap;
      sscanf(arg[0],"%hx:%hhx:%x", &nap, &uap, &lap);
   
      payload.nap = nap;
      payload.uap = uap;
      payload.lap = (lap & 0xffffff); // 24 bits.

The bluenode.lua backend also simpifies to match

local nap,uap,lap = jeenode.bitslicer(data,16,8,24)

Now we have these piece and we can rebuild the NAP:UAP:LAP (device) hex string with ease, it also means we only need to send 6 bytes, and always 6 bytes, over RF (less is better).

local device = string.format("%x:%x:%x", nap, uap, lap)

That's feels about right.

Brett

ERK - I just realized I screwed up the TTL is on ALL endpoint for a NODE and as this BlueNode now has multiple endpoint that may come and go I need to rework the expiry logic some more.

Added custom reaper on the BT endpoints for the LUA side - now we are cooking.
Gary I do try to make my code readable and commented for others to use as examples.

Brett

Hi Guys

Great work!!

I have been messing with this module on a breadboard for some time, but have a problem. When I issue the AT+INQ i get an error(16) response which according to the spec means means ‘ SPP lib hasn’t been initialized ‘. I then issue AT+INIT which is accepted and then when I issue AT+INQ the module falls out of command mode back to pairing mode. I'm a bit stuck. Any ideas?

John

Bob I found that "smart" phones are too smart for their own good these days
a) they don't leave the BT enabled when they power down.  Crap.
b) things like the iSuck's even have a requirement that its paired. wtf.

Try pulling out the oldest phone you have that has Bluetooth.  I got my best results using an 8 year old phone, strangely this is the phone I use each day as I shun the "smarts" that don't add any value except suck the live out of your battery and mess with the standards.

I does seem that all HC-05 devices are not created equal.  Mine for example didn't like the AT+INIT command.  I'll take a photo of the chips and post later.

I had another though about the bluenode.lua sketch.  It would be nice that if a device is discovered just because its not explicity mentioned as an endpoint when the BlueNode{} is setup you'd want to see it anyway.

If you register the device with a name that will be used otherwise one will be made up that will be the device.
For example:

local nodes = {
   [18] = BlueNode{instance="bluetooth:presence",
                   endpoints={
                      ["1d:fe:720479"] = "brett.1"
                }
           }
}

The device 1d:fe:720479 is setup so we get this -> dbzoo.livebox.bluetooth:presence.brett.1
As the device 12:37:e90912 is not setup we would get this -> dbzoo.livebox.bluetooth:presence.12.37.e90912

With the endpoint being dynamically created based upon its device identifier.  This means in its simpliest form you can setup this as your node and ALL your bluetooth devices will be auto discovered and appear as presence endpoints.  Woo.

   [18] = BlueNode{instance="bluetooth:presence"}

If you want to name the device as an explicit endpoint you can then use the form above...  cool.

Brett

but on the general point of smartphones shutting bluetooth down when asleep, I'm pretty convinced that I can l2ping a sleeping iPhone 4S and also a sleeping iPhone 3GS from the Raspi USB dongle setup.

I've posted up some notes http://www.dbzoo.com/livebox/hah_hahnode/bluenode

There must be a way to make ios devices respond to inquiry scans.  Aren't they permenantly discoverable?   There is a setting on most devices to enable/disable this mode.

The blue node has a reaper at the tail end of it and instead of returning the class (aka table) its returning true/false as the last function is not a class method.   I've corrected that 311.4. Strange I never hit that problem when I was testing on my dev linux box with LUA locally.  Anyway easily fixed.

Thanks for the feed back Andrew.

Brett

I'm not winning with this one
Instead of requiring like this:

BlueNode = require("xap.bluenode").BlueNode

Change it to this:

BlueNode = require("xap.bluenode")

That is because the module itself is returning the correct object you don't need to reach into and get it.

Umm. there is no blueNodeApplet.lua - that code you pasted (which I removed) is the decoder.  It should not appear in your plugboard directory at all.  Its already on your livebox in /usr/share/lua/5.1/xap/bluenode.lua

The only configuration you need is http://www.dbzoo.com/livebox/hah_hahnode/bluenode#configuration

Brett

unless you jailbreak :) in which case you dont need the apple devs license

Hi Brett,

Apologies for the late reply, I ended up crashing the livebox and had to recover with the RGWRepair tool back to 287.dwb,

i put that down to filling memory possible with a debug running in plugboard but cant be 100% sure as i also attempted your latest beta update to 311.4

Anyway that aside i am back up and running seem to be receiving a lot of invalid xap messages, my instance is set to a null, and my hostname is set to livebox

Error Message: Invalid name/value pairing. Address components cannot be empty strings.
Message Line#: 7 - source=dbzoo..jeenode:room1.light
Received From: 192.168.1.12:3077

xap-header
{
v=12
hop=1
uid=FF00D801
class=xAPBSC.info
source=dbzoo..jeenode:room1.light
}
input.state
{
state=?
text=?
}

Error Message: Invalid name/value pairing. Address components cannot be empty strings.
Message Line#: 7 - source=dbzoo..Plugboard
Received From: 192.168.1.12:3078

xap-hbeat
{
v=12
hop=1
uid=FF00D800
class=xap-hbeat.alive
source=dbzoo..Plugboard
interval=60
port=3644
}

Also, i had to add .RoomNode to the RoomNode = require("xap.roomnode") to my jeenodeapplet.lua V4 otherwise it will fail to run

# xap-plugboard
lua: /etc/plugboard/jeenodeApplet.lua:30: attempt to call global 'RoomNode' (a table value)
stack traceback:
        /etc/plugboard/jeenodeApplet.lua:30: in main chunk
        [C]: in function 'require'
        /etc_ro_fs/plugboard/plugboard.lua:23: in function 'fun'
        /usr/share/lua/5.1/pl/tablex.lua:388: in function 'foreach'
        /etc_ro_fs/plugboard/plugboard.lua:66: in main chunk
        [C]: ?
Loading /etc/plugboard/jeenodeApplet.lua        #

Thanks

Andrew

Gary in 311.4 beta that instance= variable is deprecated and shouldn't be used anymore.   The HOSTNAME of the system is now the instance.  This was done so that the code could move platforms more easily.

As Andrews hostname is livebox that should be sufficent to produce dbzoo.<hostname>.item

The xAP packet being produced is from the jeenodeApplet.lua and this too has under gone a rework.

Historically the endpoints would be defined like this:

RoomNode{base="dbzoo.livebox.jeenode:1.sensor" ...}  etc..

But this hardcoded the word "livebox" which I want to avoid so the code can move freely cross system this was changed too.
http://www.dbzoo.com/livebox/hah_plugboard_v2#bsc  If you examine how endpoints are created there are now three variables that replace this:  deviceid, vendorid and instanceid.   99% of the time you only need to set the instanceid the others will default.

Now getting back to where the bug might be.  If you have etc/xap.d/system.ini with instance= set but not defined to any value this will BREAK your system.   Perhaps the bug is mine  Could you confirm?  If this is the case delete the instance= setting. 
I will do a check and only pick up value if its not an empty string.

A few things are changing in the system to allow this code to be more system agnostic in preparation for running on different hardware, unfortunately this mean there will be a little teething pain in moving to the beta and debugging for me.

Brett

Thanks for clearing that up Brett

i just wasnt getting the whole instance mods but as you mentioned 311.4 changes things and makes sense now

and yes you are correct, the [xap] instance needs to check for null and not process.

thanks

Going back to the bluenode / jeenode i appear to be picking up roomnode data but these are not being assinged to my endpoints

# microcom -s 57600 /dev/ttyUSB2

[RF12nocfg.1] A i1 g212 @ 868 MHz

Current basenode configuration:
 A i1 g212 @ 868 MHz

OK 2 5 58 29 0

xap-header
{
v=12
hop=1
uid=FF00D802
class=xAPBSC.info
source=dbzoo.livebox.jeenode:room1.temp
}
input.state
{
state=?
text=?
}

Andrew

I've pushed 311.5

I found a bug with having a defined but empty instance= setting, which you reported to be your problem.

The way you include the "nodules" for setting up decoders has changed in this update as well to unify how this is setup.

OLD:  RoomNode = require ("xap.roomnode").RoomNode
NEW: RoomNode = require ("xap.roomnode")

This streamlines the jeenode applet configuration with how the bluenode works.   The others should have been like this too.
The wiki has been updated to reflect the new world of the 311 beta as I like to update the docs as I make changes, but it means its in a funny transition period in that what is says won't work for those on the 311 prod release.  Hopefully this period won't be too long.

Thanks for letting me know about that bug... One more down !

If you update to 311.5 you'll be able to remove/null your instance= setting back out (really you don't want to use this).

Please bear with me during this transition period of 311->312

Thanks Brett,

updated to 311.5 and modded jeenodeapplet.lua  with "RoomNode = require ("xap.roomnode")" all seems to run fine now both automatically and manually with xap-plugboard

Yes i agree, i think thats a good idea - change [xap] instance to deviceid

i can also confirm my [xap] setting contains no settings and my hostname is being used as the deviceid so the fallback works as you stated.

just one issue now - no data from the roomnode / bluenode being mapped to the endpoints

Thanks for your efforts with the beta, looking good and the saving from gui is lightening fast!

Andrew

Just getting back in to the swing of things with the Bluenode

Im current running Bretts sketch - http://livebox-hah.googlecode.com/svn/trunk/userapps/arduino/Bluetooth/B...

i have updated the sketch from NODE_18 to NODE_ID 3 and added the bluenode references to my jeenodeapplet V4.0 as below

-- Keyed by NODE ID
local nodes = {
   [2] = RoomNode{instance="jeenode:room1", endpoints={temp=1,light=1,humi=1,moved=1,lobat=1}, ttl=900},
   [3] = BlueNode{instance="bluetooth:presence"},

From the WIKI i can see i should be receiving data resembling room node responses i.e data = "OK 18 18 0 55 18 9 233",

Do i need to flash HAH_central and configure the RFM to the correct group / id before uploading the sketch?

Also, currently looking at the debug of my jeenodeapplet and it does not appear to be reaping the bluenode

# xap-plugboard
Loading /etc/plugboard/jeenodeApplet.lua        [ JeeNode ]
build RoomNode[2]
build BlueNode[3]

Running...
Reaping RoomNode[2] alive=true ttl=900 timeout=10
Reaping RoomNode[2] alive=true ttl=900 timeout=20
Reaping RoomNode[2] alive=true ttl=900 timeout=30

thanks

Andrew

Thanks Brett,

So the process is the build a fresh room node and flash just the bluenode sketch and nothing else is required?

no other sketches and no config of the RFM module?

No xap-serial from the node as far as i can see in both xfx viewer and using microm

Debugging output

[BlueNode.1]3 g212 @ 868Mhz
SYNC> AT
OK
SYNC> AT+NAME=blueNode
OK
SYNC> AT+ROLE=1
OK
SYNC> AT+INQM=0,5,15
OK
ASYNC> AT+INQ
ERROR:(16)
ASYNC> AT+INQ
ERROR:(16)

Thanks

Andrew

thanks Brett,

Will give it a test this evening after work

Andrew

Andrew from that output it looks like the AT+INQ command is working, strange about the error 17 thou.
As you say its not finding any BT devices to discover.  If you have any non-apple products I would try to discover those first.

Brett

Tried a number of different handsets

Lots of

ASYNC> AT+INQ
OK
ASYNC> AT+INQ
OK

but no devices detected, i've also tested with the BlueTerm sketch but this seems to freeze before i can send any AT commands, will post some debug later from Arduino

thanks

Andrew