Refactor to make some real design docs

docs/software/bluetooth-api.md

@@ -0,0 +1,111 @@
+# Bluetooth API
+
+The Bluetooth API is design to have only a few characteristics and most polymorphism comes from the flexible set of Google Protocol Buffers which are sent over the wire.  We use protocol buffers extensively both for the bluetooth API and for packets inside the mesh or when providing packets to other applications on the phone.
+
+## A note on MTU sizes
+
+This device will work with any MTU size, but it is highly recommended that you call your phone's "setMTU function to increase MTU to 512 bytes" as soon as you connect to a service.  This will dramatically improve performance when reading/writing packets.
+
+## MeshBluetoothService 
+
+This is the main bluetooth service for the device and provides the API your app should use to get information about the mesh, send packets or provision the radio.  
+
+For a reference implementation of a client that uses this service see [RadioInterfaceService](https://github.com/meshtastic/Meshtastic-Android/blob/master/app/src/main/java/com/geeksville/mesh/service/RadioInterfaceService.kt).  Typical flow when 
+a phone connects to the device should be the following:
+
+* SetMTU size to 512
+* Read a RadioConfig from "radio" - used to get the channel and radio settings
+* Read (and write if incorrect) a User from "user" - to get the username for this node
+* Read a MyNodeInfo from "mynode" to get information about this local device
+* Write an empty record to "nodeinfo" to restart the nodeinfo reading state machine
+* Read from "nodeinfo" until it returns empty to build the phone's copy of the current NodeDB for the mesh
+* Read from "fromradio" until it returns empty to get any messages that arrived for this node while the phone was away
+* Subscribe to notify on "fromnum" to get notified whenever the device has a new received packet
+* Read that new packet from "fromradio"
+* Whenever the phone has a packet to send write to "toradio"
+
+For definitions (and documentation) of FromRadio, ToRadio, MyNodeInfo, NodeInfo and User  protocol buffers see [mesh.proto](https://github.com/meshtastic/Meshtastic-protobufs/blob/master/mesh.proto)
+
+UUID for the service: 6ba1b218-15a8-461f-9fa8-5dcae273eafd
+
+Each characteristic is listed as follows:
+
+UUID
+Properties
+Description (including human readable name)
+
+8ba2bcc2-ee02-4a55-a531-c525c5e454d5
+read
+fromradio - contains a newly received FromRadio packet destined towards the phone (up to MAXPACKET bytes per packet).
+After reading the esp32 will put the next packet in this mailbox.  If the FIFO is empty it will put an empty packet in this
+mailbox.
+
+f75c76d2-129e-4dad-a1dd-7866124401e7
+write
+toradio - write ToRadio protobufs to this characteristic to send them (up to MAXPACKET len)
+
+ed9da18c-a800-4f66-a670-aa7547e34453
+read,notify,write
+fromnum - the current packet # in the message waiting inside fromradio, if the phone sees this notify it should read messages
+until it catches up with this number.
+
+The phone can write to this register to go backwards up to FIXME packets, to handle the rare case of a fromradio packet was dropped after the esp32 callback was called, but before it arrives at the phone.  If the phone writes to this register the esp32 will discard older packets and put the next packet >= fromnum in fromradio.
+When the esp32 advances fromnum, it will delay doing the notify by 100ms, in the hopes that the notify will never actally need to be sent if the phone is already pulling from fromradio.
+
+Note: that if the phone ever sees this number decrease, it means the esp32 has rebooted.
+
+ea9f3f82-8dc4-4733-9452-1f6da28892a2
+read
+mynode - read this to access a MyNodeInfo protobuf
+
+d31e02e0-c8ab-4d3f-9cc9-0b8466bdabe8
+read, write
+nodeinfo - read this to get a series of NodeInfos (ending with a null empty record), write to this to restart the read statemachine that returns all the node infos
+
+b56786c8-839a-44a1-b98e-a1724c4a0262
+read,write
+radio - read/write this to access a RadioConfig protobuf
+
+6ff1d8b6-e2de-41e3-8c0b-8fa384f64eb6
+read,write
+owner - read/write this to access a User protobuf
+
+Re: queue management
+Not all messages are kept in the fromradio queue (filtered based on SubPacket):
+* only the most recent Position and User messages for a particular node are kept
+* all Data SubPackets are kept
+* No WantNodeNum / DenyNodeNum messages are kept
+A variable keepAllPackets, if set to true will suppress this behavior and instead keep everything for forwarding to the phone (for debugging)
+
+
+## Other bluetooth services
+
+This document focuses on the core mesh service, but it is worth noting that the following other Bluetooth services are also
+provided by the device.
+
+### BluetoothSoftwareUpdate
+
+The software update service.  For a sample function that performs a software update using this API see [startUpdate](https://github.com/meshtastic/Meshtastic-Android/blob/master/app/src/main/java/com/geeksville/mesh/service/SoftwareUpdateService.kt).
+
+SoftwareUpdateService UUID cb0b9a0b-a84c-4c0d-bdbb-442e3144ee30
+
+Characteristics
+
+| UUID                                 | properties       | description|
+|--------------------------------------|------------------|------------|
+
+| e74dd9c0-a301-4a6f-95a1-f0e1dbea8e1e | write,read       | total image size, 32 bit, write this first, then read read back to see if it was acceptable (0 mean not accepted) |
+| e272ebac-d463-4b98-bc84-5cc1a39ee517 | write            | data, variable sized, recommended 512 bytes, write one for each block of file |
+| 4826129c-c22a-43a3-b066-ce8f0d5bacc6 | write            | crc32, write last - writing this will complete the OTA operation, now you can read result |
+| 5e134862-7411-4424-ac4a-210937432c77 | read,notify      | result code, readable but will notify when the OTA operation completes |
+| GATT_UUID_SW_VERSION_STR/0x2a28 | read | We also implement these standard GATT entries because SW update probably needs them: |
+| GATT_UUID_MANU_NAME/0x2a29 | read | |
+| GATT_UUID_HW_VERSION_STR/0x2a27 | read | |
+
+### DeviceInformationService
+
+Implements the standard BLE contract for this service (has software version, hardware model, serial number, etc...)
+
+### BatteryLevelService
+
+Implements the standard BLE contract service, provides battery level in a way that most client devices should automatically understand (i.e. it should show in the bluetooth devices screen automatically)

docs/software/mesh-alg.md

@@ -0,0 +1,66 @@
+# Mesh broadcast algorithm
+
+FIXME - instead look for standard solutions.  this approach seems really suboptimal, because too many nodes will try to rebroast.  If
+all else fails could always use the stock Radiohead solution - though super inefficient.
+
+TODO:
+* reread the radiohead mesh implementation
+* read about general mesh flooding solutions
+* reread the disaster radio protocol docs
+
+good description of batman protocol: https://www.open-mesh.org/projects/open-mesh/wiki/BATMANConcept
+
+interesting paper on lora mesh: https://portal.research.lu.se/portal/files/45735775/paper.pdf
+It seems like  DSR might be the algorithm used by RadioheadMesh.  DSR is described in https://tools.ietf.org/html/rfc4728
+https://en.wikipedia.org/wiki/Dynamic_Source_Routing
+
+broadcast solution:
+Use naive flooding at first (FIXME - do some math for a 20 node, 3 hop mesh.  A single flood will require a max of 20 messages sent)
+Then move to MPR later (http://www.olsr.org/docs/report_html/node28.html).  Use altitude and location as heursitics in selecting the MPR set
+
+compare to db sync algorithm?
+
+what about never flooding gps broadcasts.  instead only have them go one hop in the common case, but if any node X is looking at the position of Y on their gui, then send a unicast to Y asking for position update.  Y replies.
+
+If Y were to die, at least the neighbor nodes of Y would have their last known position of Y.
+
+## approach 1
+
+* send all broadcasts with a TTL
+* periodically(?) do a survey to find the max TTL that is needed to fully cover the current network.
+* to do a study first send a broadcast (maybe our current initial user announcement?) with TTL set to one (so therefore no one will rebroadcast our request)
+* survey replies are sent unicast back to us (and intervening nodes will need to keep the route table that they have built up based on past packets)
+* count the number of replies to this TTL 1 attempt.  That is the number of nodes we can reach without any rebroadcasts
+* repeat the study with a TTL of 2 and then 3.  stop once the # of replies stops going up.
+* it is important for any node to do listen before talk to prevent stomping on other rebroadcasters...
+* For these little networks I bet a max TTL would never be higher than 3?
+
+## approach 2
+
+* send a TTL1 broadcast, the replies let us build a list of the nodes (stored as a bitvector?) that we can see (and their rssis)
+* we then broadcast out that bitvector (also TTL1) asking "can any of ya'll (even indirectly) see anyone else?"
+* if a node can see someone I missed (and they are the best person to see that node), they reply (unidirectionally) with the missing nodes and their rssis (other nodes might sniff (and update their db) based on this reply but they don't have to)
+* given that the max number of nodes in this mesh will be like 20 (for normal cases), I bet globally updating this db of "nodenums and who has the best rssi for packets from that node" would be useful
+* once the global DB is shared, when a node wants to broadcast, it just sends out its broadcast . the first level receivers then make a decision "am I the best to rebroadcast to someone who likely missed this packet?" if so, rebroadcast
+
+## approach 3
+
+* when a node X wants to know other nodes positions, it broadcasts its position with want_replies=true.  Then each of the nodes that received that request broadcast their replies (possibly by using special timeslots?)
+* all nodes constantly update their local db based on replies they witnessed.
+* after 10s (or whatever) if node Y notices that it didn't hear a reply from node Z (that Y has heard from recently ) to that initial request, that means Z never heard the request from X.  Node Y will reply to X on Z's behalf.
+* could this work for more than one hop?  Is more than one hop needed?  Could it work for sending messages (i.e. for a msg sent to Z with want-reply set). 
+
+## approach 4
+
+look into the literature for this idea specifically.
+
+* don't view it as a mesh protocol as much as a "distributed db unification problem".  When nodes talk to nearby nodes they work together
+to update their nodedbs.  Each nodedb would have a last change date and any new changes that only one node has would get passed to the 
+other node.  This would nicely allow distant nodes to propogate their position to all other nodes (eventually).
+* handle group messages the same way, there would be a table of messages and time of creation.
+* when a node has a new position or message to send out, it does a broadcast.  All the adjacent nodes update their db instantly (this handles 90% of messages I'll bet).  
+* Occasionally a node might broadcast saying "anyone have anything newer than time X?"  If someone does, they send the diffs since that date.
+* essentially everything in this variant becomes broadcasts of "request db updates for >time X - for _all_ or for a particular nodenum" and nodes sending (either due to request or because they changed state) "here's a set of db updates".  Every node is constantly trying to
+build the most recent version of reality, and if some nodes are too far, then nodes closer in will eventually forward their changes to the distributed db.
+* construct non ambigious rules for who broadcasts to request db updates.  ideally the algorithm should nicely realize node X can see most other nodes, so they should just listen to all those nodes and minimize the # of broadcasts. the distributed picture of nodes rssi could be useful here?
+* possibly view the BLE protocol to the radio the same way - just a process of reconverging the node/msgdb database.

docs/software/power.md

@@ -0,0 +1,81 @@
+# Power Management State Machine
+
+i.e. sleep behavior
+
+## States
+
+From lower to higher power consumption.
+
+* Super-deep-sleep (SDS) - everything is off, CPU, radio, bluetooth, GPS.  Only wakes due to timer or button press.  We enter this mode only after no radio comms for a few hours, used to put the device into what is effectively "off" mode.
+  onEntry: setBluetoothOn(false), call doDeepSleep
+  onExit: (standard bootup code, starts in DARK)
+
+* deep-sleep (DS) - CPU is off, radio is on, bluetooth and GPS is off.  Note: This mode is never used currently, because it only saves 1.5mA vs light-sleep
+  (Not currently used)
+
+* light-sleep (LS) - CPU is suspended (RAM stays alive), radio is on, bluetooth is off, GPS is off.  Note: currently GPS is not turned 
+off during light sleep, but there is a TODO item to fix this.
+  onEntry: setBluetoothOn(false), setGPSPower(false), doLightSleep()
+  onIdle: (if we wake because our led blink timer has expired) blink the led then go back to sleep until we sleep for ls_secs
+  onExit: setGPSPower(true), start trying to get gps lock: gps.startLock(), once lock arrives service.sendPosition(BROADCAST)
+
+* No bluetooth (NB) - CPU is running, radio is on, GPS is on but bluetooth is off, screen is off.
+  onEntry: setBluetoothOn(false)
+  onExit: 
+
+* running dark (DARK) - Everything is on except screen
+  onEntry: setBluetoothOn(true)
+  onExit: 
+
+* full on (ON) - Everything is on
+  onEntry: setBluetoothOn(true), screen.setOn(true)
+  onExit: screen.setOn(false)
+
+## Behavior
+
+### events that increase CPU activity
+
+* At cold boot: The initial state (after setup() has run) is DARK
+* While in DARK: if we receive EVENT_BOOT, transition to ON (and show the bootscreen).  This event will be sent if we detect we woke due to reset (as opposed to deep sleep)
+* While in LS: Once every position_broadcast_secs (default 15 mins) - the unit will wake into DARK mode and broadcast a "networkPing" (our position) and stay alive for wait_bluetooth_secs (default 30 seconds).  This allows other nodes to have a record of our last known position if we go away and allows a paired phone to hear from us and download messages.
+* While in LS: Every send_owner_interval (defaults to 4, i.e. one hour), when we wake to send our position we _also_ broadcast our owner.  This lets new nodes on the network find out about us or correct duplicate node number assignments.
+* While in LS/NB/DARK: If the user presses a button (EVENT_PRESS) we go to full ON mode for screen_on_secs (default 30 seconds).  Multiple presses keeps resetting this timeout
+* While in LS/NB/DARK: If we receive new text messages (EVENT_RECEIVED_TEXT_MSG), we go to full ON mode for screen_on_secs (same as if user pressed a button)
+* While in LS: while we receive packets on the radio (EVENT_RECEIVED_PACKET) we will wake and handle them and stay awake in NB mode for min_wake_secs (default 10 seconds) 
+* While in NB: If we do have packets the phone (EVENT_PACKETS_FOR_PHONE) would want we transition to DARK mode for wait_bluetooth secs.
+* While in DARK/ON: If we receive EVENT_BLUETOOTH_PAIR we transition to ON and start our screen_on_secs timeout
+* While in NB/DARK/ON: If we receive EVENT_NODEDB_UPDATED we transition to ON (so the new screen can be shown)
+* While in DARK: While the phone talks to us over BLE (EVENT_CONTACT_FROM_PHONE) reset any sleep timers and stay in DARK (needed for bluetooth sw update and nice user experience if the user is reading/replying to texts)
+
+### events that decrease cpu activity
+
+* While in ON: If PRESS event occurs, reset screen_on_secs timer and tell the screen to handle the pess
+* While in ON: If it has been more than screen_on_secs since a press, lower to DARK
+* While in DARK: If time since last contact by our phone exceeds phone_timeout_secs (15 minutes), we transition down into NB mode
+* While in DARK or NB: If nothing above is forcing us to stay in a higher mode (wait_bluetooth_secs, min_wake_secs) we will lower down to LS state
+* While in LS: If either phone_sds_timeout_secs (default 2 hr) or mesh_sds_timeout_secs (default 2 hr) are exceeded we will lower into SDS mode for sds_secs (default 1 yr) (or a button press).  (Note: phone_sds_timeout_secs is currently disabled for now, because most users 
+are using without a phone)
+* Any time we enter LS mode: We stay in that until an interrupt, button press or other state transition.  Every ls_secs (default 1 hr) and let the arduino loop() run one iteration (FIXME, not sure if we need this at all), and then immediately reenter lightsleep mode on the CPU.
+
+TODO: Eventually these scheduled intervals should be synchronized to the GPS clock, so that we can consider leaving the lora receiver off to save even more power.
+TODO: In NB mode we should put cpu into light sleep any time we really aren't that busy (without declaring LS state) - i.e. we should leave GPS on etc...
+
+# Low power consumption tasks
+
+General ideas to hit the power draws our spreadsheet predicts.  Do the easy ones before beta, the last 15% can be done after 1.0.
+
+* don't even power on the gps until someone else wants our position, just stay in lora deep sleep until press or rxpacket (except for once an hour updates)
+* (possibly bad idea - better to have lora radio always listen - check spreadsheet) have every node wake at the same tick and do their position syncs then go back to deep sleep
+* lower BT announce interval to save battery
+* change to use RXcontinuous mode and config to drop packets with bad CRC (see section 6.4 of datasheet) - I think this is already the case
+* have mesh service run in a thread that stays blocked until a packet arrives from the RF95
+* platformio sdkconfig CONFIG_PM and turn on modem sleep mode
+* keep cpu 100% in deepsleep until irq from radio wakes it.  Then stay awake for 30 secs to attempt delivery to phone.  
+* use https://lastminuteengineers.com/esp32-sleep-modes-power-consumption/ association sleep pattern to save power - but see https://github.com/espressif/esp-idf/issues/2070 and https://esp32.com/viewtopic.php?f=13&t=12182 it seems with BLE on the 'easy' draw people are getting is 80mA
+* stop using loop() instead use a job queue and let cpu sleep
+* measure power consumption and calculate battery life assuming no deep sleep
+* do lowest sleep level possible where BT still works during normal sleeping, make sure cpu stays in that mode unless lora rx packet happens, bt rx packet happens or button press happens
+* optionally do lora messaging only during special scheduled intervals (unless nodes are told to go to low latency mode), then deep sleep except during those intervals - before implementing calculate what battery life would be with  this feature
+* see section 7.3 of https://cdn.sparkfun.com/assets/learn_tutorials/8/0/4/RFM95_96_97_98W.pdf and have hope radio wake only when a valid packet is received.  Possibly even wake the ESP32 from deep sleep via GPIO.
+* never enter deep sleep while connected to USB power (but still go to other low power modes)
+* when main cpu is idle (in loop), turn cpu clock rate down and/or activate special sleep modes.  We want almost everything shutdown until it gets an interrupt.

docs/software/sw-design.md

@@ -0,0 +1,6 @@
+This is a mini design doc for various core behaviors...
+
+* [Power Management](power.md)
+* [Mesh algorithm](mesh-alg.md)
+* [Bluetooth API](bluetooth-api.md) and porting guide for new clients (iOS, python, etc...)
+* TODO: how to port the device code to a new device.