mbed-os/features/lorawan/LoRaWANStack.h

502 lines
21 KiB
C
Raw Normal View History

/**
/ _____) _ | |
( (____ _____ ____ _| |_ _____ ____| |__
\____ \| ___ | (_ _) ___ |/ ___) _ \
_____) ) ____| | | || |_| ____( (___| | | |
(______/|_____)_|_|_| \__)_____)\____)_| |_|
(C)2013 Semtech
___ _____ _ ___ _ _____ ___ ___ ___ ___
/ __|_ _/_\ / __| |/ / __/ _ \| _ \/ __| __|
\__ \ | |/ _ \ (__| ' <| _| (_) | / (__| _|
|___/ |_/_/ \_\___|_|\_\_| \___/|_|_\\___|___|
embedded.connectivity.solutions===============
Description: LoRaWAN stack layer that controls both MAC and PHY underneath
License: Revised BSD License, see LICENSE.TXT file include in the project
Maintainer: Miguel Luis ( Semtech ), Gregory Cristian ( Semtech ) and Daniel Jaeckle ( STACKFORCE )
Copyright (c) 2017, Arm Limited and affiliates.
SPDX-License-Identifier: BSD-3-Clause
*/
#ifndef LORAWANSTACK_H_
#define LORAWANSTACK_H_
#include <stdint.h>
#include "events/EventQueue.h"
#include "platform/Callback.h"
#include "platform/NonCopyable.h"
#include "lorawan/system/LoRaWANTimer.h"
#include "lorastack/mac/LoRaMac.h"
#include "lorawan/system/lorawan_data_structures.h"
#include "LoRaRadio.h"
/**
* A mask for the network ID.
*/
#define LORAWAN_NETWORK_ID_MASK ( uint32_t )0xFE000000
class LoRaWANStack: private mbed::NonCopyable<LoRaWANStack> {
private:
/** End-device states.
*
*/
typedef enum device_states {
DEVICE_STATE_NOT_INITIALIZED,
DEVICE_STATE_INIT,
DEVICE_STATE_JOINING,
DEVICE_STATE_ABP_CONNECTING,
DEVICE_STATE_JOINED,
DEVICE_STATE_SEND,
DEVICE_STATE_IDLE,
#if defined(LORAWAN_COMPLIANCE_TEST)
DEVICE_STATE_COMPLIANCE_TEST,
#endif
DEVICE_STATE_SHUTDOWN
} device_states_t;
public:
static LoRaWANStack& get_lorawan_stack();
/** Binds radio driver to PHY layer.
*
* MAC layer is totally detached from the PHY layer so the stack layer
* needs to play the role of an arbitrator. This API gets a radio driver
* object from the application (via LoRaWANInterface), binds it to the PHY
* layer and initialises radio callback handles which the radio driver will
* use in order to report events.
*
* @param radio LoRaRadio object, i.e., the radio driver
*
*/
void bind_radio_driver(LoRaRadio& radio);
/** Connect OTAA or ABP using Mbed-OS config system
*
* Connect by Over The Air Activation or Activation By Personalization.
* You need to configure the connection properly via the Mbed OS configuration
* system.
*
* When connecting via OTAA, the return code for success (LORAWAN_STATUS_CONNECT_IN_PROGRESS) is negative.
* However, this is not a real error. It tells you that the connection is in progress and you will
* be notified of the completion via an event. By default, after the Join Accept message
* is received, base stations may provide the node with a CF-List that replaces
* all user-configured channels except the Join/Default channels. A CF-List can
* configure a maximum of five channels other than the default channels.
*
* In case of ABP, the CONNECTED event is posted before the call to `connect()` returns.
* To configure more channels, we recommend that you use the `set_channel_plan()` API after the connection.
* By default, the PHY layers configure only the mandatory Join channels. The retransmission back-off restrictions
* on these channels are severe and you may experience long delays or even failures in the confirmed traffic.
* If you add more channels, the aggregated duty cycle becomes much more relaxed as compared to the Join (default) channels only.
*
* **NOTES ON RECONNECTION:**
* Currently, the Mbed OS LoRaWAN implementation does not support non-volatile
* memory storage. Therefore, the state and frame counters cannot be restored after
* a power cycle. However, if you use the `disconnect()` API to shut down the LoRaWAN
* protocol, the state and frame counters are saved. Connecting again would try to
* restore the previous session. According to the LoRaWAN 1.0.2 specification, the frame counters are always reset
* to zero for OTAA and a new Join request lets the network server know
* that the counters need a reset. The same is said about the ABP but there
* is no way to convey this information to the network server. For a network
* server, an ABP device is always connected. That's why storing the frame counters
* is important, at least for ABP. That's why we try to restore frame counters from
* session information after a disconnection.
*
* @return LORAWAN_STATUS_OK or LORAWAN_STATUS_CONNECT_IN_PROGRESS
* on success, or a negative error code on failure.
*/
lorawan_status_t connect();
/** Connect OTAA or ABP with parameters
*
* All connection parameters are chosen by the user and provided in the
* data structure passed down.
*
* When connecting via OTAA, the return code for success (LORAWAN_STATUS_CONNECT_IN_PROGRESS) is negative.
* However, this is not a real error. It tells you that connection is in progress and you will
* be notified of completion via an event. By default, after Join Accept message
* is received, base stations may provide the node with a CF-List which replaces
* all user-configured channels except the Join/Default channels. A CF-List can
* configure a maximum of five channels other than the default channels.
*
* In case of ABP, the CONNECTED event is posted before the call to `connect()` returns.
* To configure more channels, we recommend that you use the `set_channel_plan()` API after the connection.
* By default, the PHY layers configure only the mandatory Join
* channels. The retransmission back-off restrictions on these channels
* are severe and you may experience long delays or even
* failures in the confirmed traffic. If you add more channels, the aggregated duty
* cycle becomes much more relaxed as compared to the Join (default) channels only.
*
* **NOTES ON RECONNECTION:**
* Currently, the Mbed OS LoRaWAN implementation does not support non-volatile
* memory storage. Therefore, the state and frame counters cannot be restored after
* a power cycle. However, if you use the `disconnect()` API to shut down the LoRaWAN
* protocol, the state and frame counters are saved. Connecting again would try to
* restore the previous session. According to the LoRaWAN 1.0.2 specification, the frame counters are always reset
* to zero for OTAA and a new Join request lets the network server know
* that the counters need a reset. The same is said about the ABP but there
* is no way to convey this information to the network server. For a network
* server, an ABP device is always connected. That's why storing the frame counters
* is important, at least for ABP. That's why we try to restore frame counters from
* session information after a disconnection.
*
* @param connect Options for an end device connection to the gateway.
*
* @return LORAWAN_STATUS_OK or LORAWAN_STATUS_CONNECT_IN_PROGRESS,
* a negative error code on failure.
*/
lorawan_status_t connect(const lorawan_connect_t &connect);
/** End device initialization.
* @param queue A pointer to an EventQueue passed from the application.
* @return LORAWAN_STATUS_OK on success, a negative error code on failure.
*/
lorawan_status_t initialize_mac_layer(events::EventQueue *queue);
/** Sets all callbacks for the application.
*
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
* @param callbacks A pointer to the structure carrying callbacks.
* @return LORAWAN_STATUS_OK on success, a negative error code on failure.
*/
lorawan_status_t set_lora_callbacks(lorawan_app_callbacks_t *callbacks);
/** Adds channels to use.
*
* You can provide a list of channels with appropriate parameters filled
* in. However, this list is not absolute. In some regions, a CF
* list gets implemented by default, which means that the network can overwrite your channel
* frequency settings right after receiving a Join Accept. You may try
* to set up any channel or channels after that and if the channel requested
* is already active, the request is silently ignored. A negative error
* code is returned if there is any problem with parameters.
*
* You need to ensure that the base station nearby supports the channel or channels being added.
*
* If your list includes a default channel (a channel where Join Requests
* are received) you cannot fully configure the channel parameters.
* Either leave the channel settings to default or check your
* corresponding PHY layer implementation. For example, LoRaPHYE868.
*
* @param channel_plan A list of channels or a single channel.
*
* @return LORAWAN_STATUS_OK on success, a negative error
* code on failure.
*/
lorawan_status_t add_channels(const lorawan_channelplan_t &channel_plan);
/** Removes a channel from the list.
*
* @param channel_id Index of the channel being removed
*
* @return LORAWAN_STATUS_OK on success, a negative error
* code on failure.
*/
lorawan_status_t remove_a_channel(uint8_t channel_id);
/** Removes a previously set channel plan.
*
* @return LORAWAN_STATUS_OK on success, a negative error
* code on failure.
*/
lorawan_status_t drop_channel_list();
/** Gets a list of currently enabled channels .
*
* @param channel_plan The channel plan structure to store final result.
*
* @return LORAWAN_STATUS_OK on success, a negative error
* code on failure.
*/
lorawan_status_t get_enabled_channels(lorawan_channelplan_t &channel_plan);
/** Sets up a retry counter for confirmed messages.
*
* Valid only for confirmed messages. This API sets the number of times the
* stack will retry a CONFIRMED message before giving up and reporting an
* error.
*
* @param count The number of retries for confirmed messages.
*
* @return LORAWAN_STATUS_OK or a negative error code.
*/
lorawan_status_t set_confirmed_msg_retry(uint8_t count);
/** Sets up the data rate.
*
* `set_datarate()` first verifies whether the data rate given is valid or not.
* If it is valid, the system sets the given data rate to the channel.
*
* @param data_rate The intended data rate, for example DR_0 or DR_1.
* Note that the macro DR_* can mean different
* things in different regions.
*
* @return LORAWAN_STATUS_OK if everything goes well, otherwise
* a negative error code.
*/
lorawan_status_t set_channel_data_rate(uint8_t data_rate);
/** Enables ADR.
*
* @param adr_enabled 0 ADR disabled, 1 ADR enabled.
*
* @return LORAWAN_STATUS_OK on success, a negative error
* code on failure.
*/
lorawan_status_t enable_adaptive_datarate(bool adr_enabled);
/** Send message to gateway
*
* @param port The application port number. Port numbers 0 and 224
* are reserved, whereas port numbers from 1 to 223
* (0x01 to 0xDF) are valid port numbers.
* Anything out of this range is illegal.
*
* @param data A pointer to the data being sent. The ownership of the
* buffer is not transferred. The data is copied to the
* internal buffers.
*
* @param length The size of data in bytes.
*
* @param flags A flag used to determine what type of
* message is being sent, for example:
*
* MSG_UNCONFIRMED_FLAG = 0x01
* MSG_CONFIRMED_FLAG = 0x02
* MSG_MULTICAST_FLAG = 0x04
* MSG_PROPRIETARY_FLAG = 0x08
* MSG_MULTICAST_FLAG and MSG_PROPRIETARY_FLAG can be
* used in conjunction with MSG_UNCONFIRMED_FLAG and
* MSG_CONFIRMED_FLAG depending on the intended use.
*
* MSG_PROPRIETARY_FLAG|MSG_CONFIRMED_FLAG mask will set
* a confirmed message flag for a proprietary message.
* MSG_CONFIRMED_FLAG and MSG_UNCONFIRMED_FLAG are
* mutually exclusive.
*
* @param null_allowed Internal use only. Needed for sending empty packet
* having CONFIRMED bit on.
*
* @return The number of bytes sent, or
* LORAWAN_STATUS_WOULD_BLOCK if another TX is
* ongoing, or a negative error code on failure.
*/
int16_t handle_tx(uint8_t port, const uint8_t* data,
uint16_t length, uint8_t flags, bool null_allowed = false);
/** Receives a message from the Network Server.
*
* @param port The application port number. Port numbers 0 and 224
* are reserved, whereas port numbers from 1 to 223
* (0x01 to 0xDF) are valid port numbers.
* Anything out of this range is illegal.
*
* @param data A pointer to buffer where the received data will be
* stored.
*
* @param length The size of data in bytes
*
* @param flags A flag is used to determine what type of
* message is being received, for example:
*
* MSG_UNCONFIRMED_FLAG = 0x01,
* MSG_CONFIRMED_FLAG = 0x02
* MSG_MULTICAST_FLAG = 0x04,
* MSG_PROPRIETARY_FLAG = 0x08
*
* MSG_MULTICAST_FLAG and MSG_PROPRIETARY_FLAG can be
* used in conjunction with MSG_UNCONFIRMED_FLAG and
* MSG_CONFIRMED_FLAG depending on the intended use.
*
* MSG_PROPRIETARY_FLAG|MSG_CONFIRMED_FLAG mask will set
* a confirmed message flag for a proprietary message.
*
* MSG_CONFIRMED_FLAG and MSG_UNCONFIRMED_FLAG are
* not mutually exclusive, i.e., the user can subscribe to
* receive both CONFIRMED AND UNCONFIRMED messages at
* the same time.
*
* @return It could be one of these:
* i) 0 if there is nothing else to read.
* ii) Number of bytes written to user buffer.
* iii) LORAWAN_STATUS_WOULD_BLOCK if there is
* nothing available to read at the moment.
* iv) A negative error code on failure.
*/
int16_t handle_rx(const uint8_t port, uint8_t* data,
uint16_t length, uint8_t flags);
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
/** Send Link Check Request MAC command.
*
*
* This API schedules a Link Check Request command (LinkCheckReq) for the network
* server and once the response, i.e., LinkCheckAns MAC command is received
* from the Network Server, an event is generated.
*
* A callback function for the link check response must be set prior to using
* this API, otherwise a LORAWAN_STATUS_PARAMETER_INVALID error is thrown.
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
*
* @return LORAWAN_STATUS_OK on successfully queuing a request, or
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
* a negative error code on failure.
*
*/
lorawan_status_t set_link_check_request();
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
/** Removes link check request sticky MAC command.
*
* Any already queued request may still get entertained. However, no new
* requests will be made.
*/
void remove_link_check_request();
/** Shuts down the LoRaWAN protocol.
*
* In response to the user call for disconnection, the stack shuts down itself.
*
* @return LORAWAN_STATUS_DEVICE_OFF on successfully shutdown.
*/
lorawan_status_t shutdown();
/** Change device class
*
* Change current device class.
*
* @param device_class The device class
*
* @return LORAWAN_STATUS_OK on success,
* LORAWAN_STATUS_UNSUPPORTED is requested class is not supported,
* or other negative error code if request failed.
*/
lorawan_status_t set_device_class(const device_class_t device_class);
private:
LoRaWANStack();
~LoRaWANStack();
/**
* Checks if the user provided port is valid or not
*/
bool is_port_valid(uint8_t port);
/**
* State machine for stack controller layer.
*/
lorawan_status_t lora_state_machine(device_states_t new_state);
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
/**
* Callback function for MLME indication. Mac layer calls this function once
* an MLME indication is received. This method translates Mac layer data
* structure into stack layer data structure.
*/
void mlme_indication_handler(loramac_mlme_indication_t *mlmeIndication);
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
/**
* Handles an MLME confirmation coming from the Mac layer and uses it to
* update the state for example, a Join Accept triggers an MLME confirmation,
* that eventually comes here and we take necessary steps accordingly.
*/
void mlme_confirm_handler(loramac_mlme_confirm_t *mlme_confirm);
/**
* Handles an MCPS request while attempting to hand over a packet from
* upper layers to Mac layer. For example in response to send_frame_to_mac(),
* an MCPS request is generated.
*/
lorawan_status_t mcps_request_handler(loramac_mcps_req_t *mcps_request);
/**
* Handles an MCPS confirmation coming from the Mac layer in response to an
* MCPS request. We take appropriate actions in response to the confirmation,
* e.g., letting the application know that ack was not received in case of
* a CONFIRMED message or scheduling error etc.
*/
void mcps_confirm_handler(loramac_mcps_confirm_t *mcps_confirm);
/**
* Handles an MCPS indication coming from the Mac layer, e.g., once we
* receive a packet from the Network Server, it is indicated to this handler
* and consequently this handler posts an event to the application that
* there is something available to read.
*/
void mcps_indication_handler(loramac_mcps_indication_t *mcps_indication);
/**
* Sets a MIB request, i.e., update a particular parameter etc.
*/
lorawan_status_t mib_set_request(loramac_mib_req_confirm_t *mib_set_params);
/**
* Requests the MIB to inquire about a particular parameter.
*/
lorawan_status_t mib_get_request(loramac_mib_req_confirm_t *mib_get_params);
/**
* Sets up user application port
*/
lorawan_status_t set_application_port(uint8_t port);
/** End device OTAA join.
*
* Based on the LoRaWAN standard 1.0.2.
* Join the network using the Over The Air Activation (OTAA) procedure.
*
* @param params The `lorawan_connect_t` type structure.
*
* @return LORAWAN_STATUS_OK or
* LORAWAN_STATUS_CONNECT_IN_PROGRESS on success,
* or a negative error code on failure.
*/
lorawan_status_t join_request_by_otaa(const lorawan_connect_t &params);
/** End device ABP join.
*
* Based on the LoRaWAN standard 1.0.2.
* Join the network using the Activation By Personalization (ABP) procedure.
*
* @param params The `lorawan_connect_t` type structure.
*
* @return LORAWAN_STATUS_OK or
* LORAWAN_STATUS_CONNECT_IN_PROGRESS on success,
* or a negative error code on failure.
*/
lorawan_status_t activation_by_personalization(const lorawan_connect_t &params);
private:
LoRaMac _loramac;
loramac_primitives_t LoRaMacPrimitives;
device_states_t _device_current_state;
lorawan_app_callbacks_t _callbacks;
lorawan_session_t _lw_session;
loramac_tx_message_t _tx_msg;
loramac_rx_message_t _rx_msg;
uint8_t _num_retry;
uint8_t _app_port;
bool _link_check_requested;
events::EventQueue *_queue;
Architecture rework, bug fixing & missing features MAC layer is now a class rather than being a blob. In addition to that Mac commands are now being handled in a seperate subsystem (a class of its own). In future we will do the same with othe sublayers of MAC like MLME, MCPS etc. The drive behind this exercise is to make MAC and supporting layers into an object oriented system. Major bug fixes include: - last join time inclusion in band parameters - disabling rx2 window if we missed the slot already - MLME uplink schdule hook - nbRep according to spec - maintaining datarate after successful joining - suppressing MLME requests if MAC is in TX_DELAYED state - Uplink dwell time verification Some missing features are implemented. Details are as follows. Support for LinkCheckRequet: An application API is added, add_link_check_request() to delegate a request for Link Check Request MAC command. * Application provides a callback function that needs to be called on reception of link check response. * Mac command is piggybacked with data frames. This API makes the sticky MAC command stick with the application payloads until/unless the application un-sticks the said mac command using remove_link_check_request() API. Handling fPending bit: If in the Downlink, we get the fPending bit set in fctrl octet, we attempt to send an empty message back to Network Server to open additional Receive windows. This operation is independent of the application. An RX_DONE event is queued bedore generating the said empty message. Specification does not mention what can be the type of that empty message. We have decided it to be of CONFIRMED type as it gives us an added benefit of retries if the corresponding RX slots are missed. Radio event callbacks as Mbed callbacks: radio_events_t structure has been carrying C-style callbacks which was inherited from the legacy code. These callbacks has now been changed to Mbed Callbacks that makes sure that we can be object oriented from now on.
2017-12-15 10:30:40 +00:00
#if defined(LORAWAN_COMPLIANCE_TEST)
/**
* Used only for compliance testing
*/
void compliance_test_handler(loramac_mcps_indication_t *mcps_indication);
/**
* Used only for compliance testing
*/
lorawan_status_t send_compliance_test_frame_to_mac();
uint8_t compliance_test_buffer[MBED_CONF_LORA_TX_MAX_SIZE];
compliance_test_t _compliance_test;
#endif
};
#endif /* LORAWANSTACK_H_ */