documentation overview for securitymanager

pull/6188/head
paul-szczepanek-arm 2018-02-28 17:00:56 +00:00
parent 701c18b428
commit d20c4efbf3
1 changed files with 94 additions and 0 deletions

View File

@ -23,6 +23,100 @@
#include "CallChainOfFunctionPointersWithContext.h"
#include "ble/BLETypes.h"
/**
* Overview
*
* Security Manager is used to provide link security through encryption, signing, and authentication
* which are made possible by pairing and optionally bonding. Pairing is the process of establishing
* and/or exchanging keys used for the current connection. Bonding means saving this information so that
* it can later be used after reconnecting without having to pair again. This saves time and power.
*
* There are many ways to provide these at different levels of security depending on your requirements
* and the facilities provided by the application. The process starts with initialising the SecurityManager
* with default options for new connections. Some settings can later be changed per link or globally.
*
* The important settings in the init() function are the MITM requirement and IO capabilities. Man in the
* Middle (MITM) protection prevents an attack where one device can impersonate another device by
* pairing with both devices at the same time. This protection is achieved sharing some information
* between the devices through some independent channel. The IO capabilities of both devices dictate
* what algorithm is used. For details @see BLUETOOTH SPECIFICATION Version 5.0 | Vol 3, Part H - 2.3.5.1.
* You can change the IO capabilities after initialisation with setIoCapability(). This will take effect
* for all subsequent pairings.
*
* Sharing this information through IO capabilities means user interaction which limits the degree of
* protection due to the limit of the amount of data that we can expect the user to transfer. Another
* solution is using OOB (out of band) communication to transfer this data instead which can send much
* more data making MITM attack even less likely to succeed. OOB data has to be exchanged by the application
* and provided to the Security Manager. Use setOOBDataUsage() to indicate you want to use it. The same call also
* allows you to set whether or not the communication channel you are using to transmit the OOB data is
* itself secure against MITM protection - this will set the level of the link security achieved using pairing
* using this data.
*
* The most secure pairing is provided by Secure Connections which relies on public key cryptography.
* Support for Secure Connections is dependent on both the stack and controller on both sides supporting
* it. If either side doesn't support it Legacy Pairing will be used. This is an older standard of pairing.
* If higher security is required legacy pairing can be disabled by calling allowLegacyPairing(false);
*
* How to use
*
* First thing you need to do is to initialise the manager by calling init() with your chosen settings.
*
* The SecurityManager communicates with your application through events. These will trigger calls in
* the EventHandler you must provide by calling the setSecurityManagerEventHandler() function.
*
* The most important process is pairing. This may be triggered manually by calling requestPairing() or
* may be called as a result of the application requiring encryption or encryption through
* requestAuthentication() or setLinkEncryption().
*
* All these can be implicitly called by useing setLinkSecurity() to conveniently set the required
* security for the link. The SecurityManager will trigger all the process required to achieve the set
* security level.
*
* Depending on the IO capabilities and OOB usage settings different pairing algorithms will be chosen.
* They will produce appropriate events which must be handled by your EventHandler.
*
* The simplest example would be a pairing of a device with no IO capabilities and no OOB data available.
* With such limited pairing capabilities the "just works" method will be employed. This does not provide
* any MITM protection. The pairing (triggered implicitly or called explicitly) will result in an event
* being generated on the peer calling pairingRequest(). The event handler must make a decision (either in
* the application itself or based on user interaction) whether to accept the pairing and call
* accetPairing() or cancelPairing(). The result will be communicated on both peers through an event calling
* pairingResult() in the EventHandler.
*
*
* Sequence diagram "Just Works" pairing
*
* /----------- Device 1 --------------\ *------ BLE link ------* /-------------- Device 2 -------------\
*
* App EventHandler SecurityManager SecurityManager EventHandler App
* | | | | | |
* |---------------------------> requestPairing() | | |
* | | |------[pairing start]------>| | |
* | | | |----------------> pairingRequest() ->|
* | | | acceptPairing() <------------------------- |
* | | |<---[pairing complete]----->| | |
* |<- pairingResult() <---------------| |----------------> pairingResult() -->|
* | | | | | |
*
* @note the requestPairing() call isn't required to trigger pairing. Pairing will also be triggered
* if you request encryption and authentication and no bonding information is available. The sequence will
* be the same save for the lack of explicit requestPairing() call.
*
*
* Sequence diagram Encryption request when bonding information is available
*
* /----------- Device 1 --------------\ *------ BLE link ------* /-------------- Device 2 -------------\
*
* App EventHandler SecurityManager SecurityManager EventHandler App
* | | | | | |
* |---------------------------> setLinkEncryption() | | |
* | | |<-[encryption established]->| | |
* |<- linkEncryptionResult() <--------| |---------> linkEncryptionResult() -->|
* | | | | | |
*
* @note if bonding information is not available, pairing will be triggered
*/
class SecurityManager {
public:
/** events sent and received when passkey is being entered */