Amazon Kinesis Video Streams WebRTC SDK for JavaScript

What is Amazon Kinesis Video Streams with WebRTC?

Amazon Kinesis Video Streams provides a standards-compliant WebRTC implementation as a fully managed capability. You can use Amazon Kinesis Video Streams with WebRTC to securely live stream media or perform two-way audio or video interaction between any camera IoT device and WebRTC-compliant mobile or web players. As a fully managed capability, you don't have to build, operate, or scale any WebRTC-related cloud infrastructure, such as signaling or media relay servers to securely stream media across applications and devices.

For more information about Amazon Kinesis Video Streams with WebRTC, check out the documentation.

This SDK is intended to be used along side the AWS SDK for JS (version 2.585.0+) to interface with the Amazon Kinesis Video Streams Signaling Service for WebRTC streaming.

Note: If you are looking to stream media from a Kinesis Video Stream (different from a Kinesis Video Streams Signaling Channel), check out the Kinesis Video Streams web media viewer.

Installing

In the Browser

To use the SDK in the browser, simply add the following script tag to your HTML pages:

1<script src="https://unpkg.com/amazon-kinesis-video-streams-webrtc/dist/kvs-webrtc.min.js"></script>

The SDK classes are made available in the global window under the KVSWebRTC namespace. For example, window.KVSWebRTC.SignalingClient.

The SDK is also compatible with bundlers like Webpack. Follow the instructions in the next section to install the NodeJS module version for use with your bundler.

In NodeJS

The preferred way to install the SDK for NodeJS is to use the npm package manager. Simply type the following into a terminal window:

1npm install amazon-kinesis-video-streams-webrtc

The SDK classes can then be imported like typical NodeJS modules:

1// JavaScript
2const SignalingClient = require('amazon-kinesis-video-streams-webrtc').SignalingClient;
3
4// TypeScript / ES6 supported
5import { SignalingClient } from 'amazon-kinesis-video-streams-webrtc';

Getting Started

You can start by trying out the SDK with a webcam on the example WebRTC test page.

It is also recommended to develop familiarity with the WebRTC protocols and KVS Signaling Channel APIs. See the following resources:

• KVS WebRTC Developer Guide
• KVS API Reference Guide

The first step in using the SDK in your own application is to follow the Installing instructions above to install the SDK.

From there, see the Usage section below for guidance on using the SDK to build a WebRTC application. Also refer to the examples directory for examples on how to write an end-to-end WebRTC application that uses the SDK.

Usage

This section demonstrates how to use this SDK along with the AWS SDK for JS (version 2.585.0+) to build a web-based viewer application. Refer to the examples directory for an example of a complete application including both a master and viewer role.

Viewer Example With Audio/Video From Local Webcam

These code snippets demonstrate how to build a viewer application that receives audio and video and also sends audio and video from a webcam back to the master.

Set Up Variables

1// DescribeSignalingChannel API can also be used to get the ARN from a channel name.
2const channelARN = 'arn:aws:kinesisvideo:us-west-2:123456789012:channel/test-channel/1234567890';
3
4// AWS Credentials
5const accessKeyId = 'ACCESS_KEY_ID_GOES_HERE';
6const secretAccessKey = 'SECRET_ACCESS_KEY_GOES_HERE';
7
8// <video> HTML elements to use to display the local webcam stream and remote stream from the master
9const localView = document.getElementsByTagName('video')[0];
10const remoteView = document.getElementsByTagName('video')[1];
11
12const region = 'us-west-2';
13const clientId = 'RANDOM_VALUE';

See Managing Credentials for more information about managing credentials in a web environment.

Create KVS Client

1const kinesisVideoClient = new AWS.KinesisVideo({
2    region,
3    accessKeyId,
4    secretAccessKey,
5    correctClockSkew: true,
6});

Get Signaling Channel Endpoints

Each signaling channel is assigned an HTTPS and WSS endpoint to connect to for data-plane operations. These can be discovered using the GetSignalingChannelEndpoint API.

1const getSignalingChannelEndpointResponse = await kinesisVideoClient
2    .getSignalingChannelEndpoint({
3        ChannelARN: channelARN,
4        SingleMasterChannelEndpointConfiguration: {
5            Protocols: ['WSS', 'HTTPS'],
6            Role: KVSWebRTC.Role.VIEWER,
7        },
8    })
9    .promise();
10const endpointsByProtocol = getSignalingChannelEndpointResponse.ResourceEndpointList.reduce((endpoints, endpoint) => {
11    endpoints[endpoint.Protocol] = endpoint.ResourceEndpoint;
12    return endpoints;
13}, {});

Create KVS Signaling Client

The HTTPS endpoint from the GetSignalingChannelEndpoint response is used with this client. This client is just used for getting ICE servers, not for actual signaling.

1const kinesisVideoSignalingChannelsClient = new AWS.KinesisVideoSignalingChannels({
2    region,
3    accessKeyId,
4    secretAccessKey,
5    endpoint: endpointsByProtocol.HTTPS,
6    correctClockSkew: true,
7});

Get ICE server configuration

For best performance, we collect STUN and TURN ICE server configurations. The KVS STUN endpoint is always stun:stun.kinesisvideo.${region}.amazonaws.com:443. To get TURN servers, the GetIceServerConfig API is used.

1const getIceServerConfigResponse = await kinesisVideoSignalingChannelsClient
2    .getIceServerConfig({
3        ChannelARN: channelARN,
4    })
5    .promise();
6const iceServers = [
7    { urls: `stun:stun.kinesisvideo.${region}.amazonaws.com:443` }
8];
9getIceServerConfigResponse.IceServerList.forEach(iceServer =>
10    iceServers.push({
11        urls: iceServer.Uris,
12        username: iceServer.Username,
13        credential: iceServer.Password,
14    }),
15);

Create RTCPeerConnection

The RTCPeerConnection is the primary interface for WebRTC communications in the Web.

1const peerConnection = new RTCPeerConnection({ iceServers });

Create WebRTC Signaling Client

This is the actual client that is used to send messages over the signaling channel.

1signalingClient = new KVSWebRTC.SignalingClient({
2    channelARN,
3    channelEndpoint: endpointsByProtocol.WSS,
4    clientId,
5    role: KVSWebRTC.Role.VIEWER,
6    region,
7    credentials: {
8        accessKeyId,
9        secretAccessKey,
10    },
11    systemClockOffset: kinesisVideoClient.config.systemClockOffset,
12});

Add Signaling Client Event Listeners

1// Once the signaling channel connection is open, connect to the webcam and create an offer to send to the master
2signalingClient.on('open', async () => {
3    // Get a stream from the webcam, add it to the peer connection, and display it in the local view
4    try {
5        const localStream = await navigator.mediaDevices.getUserMedia({
6            video: { width: { ideal: 1280 }, height: { ideal: 720 } },
7            audio: true,
8        });
9        localStream.getTracks().forEach(track => peerConnection.addTrack(track, localStream));
10        localView.srcObject = localStream;
11    } catch (e) {
12        // Could not find webcam
13        return;
14    }
15
16    // Create an SDP offer and send it to the master
17    // If there is no concern about browser compatibility, using `addTransceiver` would be better
18    const offer = await viewer.peerConnection.createOffer({
19        offerToReceiveAudio: true,
20        offerToReceiveVideo: true,
21    });
22    await peerConnection.setLocalDescription(offer);
23    signalingClient.sendSdpOffer(viewer.peerConnection.localDescription);
24});
25
26// When the SDP answer is received back from the master, add it to the peer connection.
27signalingClient.on('sdpAnswer', async answer => {
28    await peerConnection.setRemoteDescription(answer);
29});
30
31// When an ICE candidate is received from the master, add it to the peer connection.
32signalingClient.on('iceCandidate', candidate => {
33    peerConnection.addIceCandidate(candidate);
34});
35
36signalingClient.on('close', () => {
37    // Handle client closures
38});
39
40signalingClient.on('error', error => {
41    // Handle client errors
42});
43

Add Peer Connection Event Listeners

1// Send any ICE candidates generated by the peer connection to the other peer
2peerConnection.addEventListener('icecandidate', ({ candidate }) => {
3    if (candidate) {
4        signalingClient.sendIceCandidate(candidate);
5    } else {
6        // No more ICE candidates will be generated
7    }
8});
9
10// As remote tracks are received, add them to the remote view
11peerConnection.addEventListener('track', event => {
12    if (remoteView.srcObject) {
13        return;
14    }
15    remoteView.srcObject = event.streams[0];
16});

Open Signaling Connection

signalingClient.open();

Documentation

This section outlines all of the classes, events, methods, and configuration options for the SDK.

Class: SignalingClient

This class is the main class for interfacing with the KVS signaling service. It extends EventEmitter.

Constructor: new SignalingClient(config)

• config {object}
  • role {Role} "MASTER" or "VIEWER".
  • channelARN {string} ARN of a channel that exists in the AWS account.
  • channelEndpoint {string} KVS Signaling Service endpoint. Should be the "WSS" endpoint from calling the GetSignalingChannel API.
  • region {string} AWS region that the channel exists in.
  • clientId {string} Identifier to uniquely identify this client when connecting to the KVS Signaling Service. Required if the role is "VIEWER". A value should not be provided if the role is "MASTER".
  • credentials {object} Must be provided unless a requestSigner is provided. See Managing Credentials.
    • accessKeyId {string} AWS access key id.
    • secretAccessKey {string} AWS secret access key.
    • sessionToken {string} Optional. AWS session token.
  • requestSigner {RequestSigner} Optional. A custom method for overriding the default SigV4 request signing.
  • systemClockOffset {number} Optional. Applies the given offset when setting the date in the SigV4 signature. See systemClockOffset and correctClockSkew properties of the AWS SDK.

Event: 'open'

Emitted when the connection to the signaling service is open.

Event: 'sdpOffer'

• sdpOffer {RTCSessionDescription} The SDP offer received from the signaling service.
• senderClientId {string} The client id of the source of the SDP offer. The value will be null if the SDP offer is from the master.

Emitted when a new SDP offer is received over the channel. Typically only a master should receive SDP offers.

Event: 'sdpAnswer'

• sdpAnswer {RTCSessionDescription} The SDP answer received from the signaling service.
• senderClientId {string} The client id of the source of the SDP answer. The value will be null if the SDP answer is from the master.

Emitted when a new SDP answer is received over the channel. Typically only a viewer should receive SDP answers.

Event: 'iceCandidate'

• iceCandidate {RTCIceCandidate} The ICE candidate received from the signaling service.
• senderClientId {string} The client id of the source of the ICE candidate. The value will be null if the ICE candidate is from the master.

Emitted when a new ICE candidate is received over the channel.

Event: 'statusResponse'

• statusResponse {statusResponse} The status response received from the signaling service.

Emitted when a statusResponse is received over the channel.

• statusResponse {object}
  • correlationId {string} correlationId of the message for which the status is meant.
  • success {boolean} Whether this statusResponse is a success or failure. (Currently, these responses are only sent on failures.)
  • errorType {optional string} A name to uniquely identify the error.
  • statusCode {optional string} HTTP status code corresponding to the nature of the response.
  • description {optional string} A string description explaining the status.

Event: 'close'

Emitted when the connection to the signaling service is closed. Even if there is an error, as long as the connection is closed, this event will be emitted.

Event: 'error'

• error {Error}

Emitted when there is an error in the client or there is an error received from the signaling service. The connection will be closed automatically.

Method: on(event, callback)

• event {string} Event name.
• callback {function} Event handler.

Binds an event handler.

Method: open()

Opens a connection to the signaling service. An error will be thrown if there is already another connection open or opening.

Method: close()

Closes the active connection to the signaling service. Nothing will happen if there is no open connection.

Method: sendSdpOffer(sdpOffer, [recipientClientId, correlationId])

• sdpOffer {RTCSessionDescription} SDP offer to send to the recipient client.
• recipientClientId {string} The id of the client to send the SDP offer to. If no id is provided, it will be sent to the master.
• correlationId {string} A unique identifier for this message. If there was an error with this message, Signaling will send a failure StatusResponse with the same correlationId.

Method: sendSdpAnswer(sdpAnswer, [recipientClientId, correlationId])

• sdpAnswer {RTCSessionDescription} SDP answer to send to the recipient client.
• recipientClientId {string} The id of the client to send the SDP answer to. If no id is provided, it will be sent to the master.
• correlationId {string} A unique identifier for this message. If there was an error with this message, Signaling will send a failure StatusResponse with the same correlationId.

Method: sendIceCandidate(iceCandidate, [recipientClientId, correlationId])

• iceCandidate {RTCIceCandidate} ICE candidate to send to the recipient client.
• recipientClientId {string} The id of the client to send the ICE candidate to. If no id is provided, it will be sent to the master.
• correlationId {string} A unique identifier for this message. If there was an error with this message, Signaling will send a failure StatusResponse with the same correlationId.

Interface: RequestSigner

Interface for signing HTTP and WebSocket requests.

Method: getSignedURL(endpoint, queryParams, [date]) => Promise<string>

• endpoint {string} The endpoint of the URL (including protocol, host, and path).
• queryParams {object} The query parameters to include in the signed URL.
• date {Date} The date that the signature is valid (+/- 5 minutes). Default: now.
• return {Promise} The signed URL.

Class: SigV4RequestSigner

This class is used to SigV4 sign requests to the signaling service. It implements RequestSigner.

This signer is unique from the signers included in the AWS SDK for JS because it supports signing WebSocket requests.

This is a useful class to use in a NodeJS backend to sign requests and send them back to a client so that the client does not need to have AWS credentials.

Constructor: new SigV4RequestSigner(region, credentials, [service])

• region {string} The region used for signing.
• credentials {Credentials} The credentials to used for signing.
• service {string} The service name used for signing. Default: kinesisvideo.

Method: getSignedURL(endpoint, queryParams, [date]) => Promise<string>

Implementation of interface method.

• Uses the SigV4 signing mechanism.
• Supports credentials with and without a session token.
• Only supports the wss:// protocol.
• Does not support specifying an expiration.

If the signer's credentials support refreshing, they will be be refreshed if necessary before signing.

Enum: Role

An enum with the following values:

• MASTER
• VIEWER

Metrics

The DQP metrics can be enabled in the sample application by checking a box before starting the viewer on the JS side. On the master, there is a flag in the sample application which (pSampleConfiguration->enableSendingMetricsToViewerViaDc) can be set to TRUE to send metrics from the master to the JS viewer. This helps get a detailed breakdown of time-to-first-frame and all the processes and API calls on master and the viewer both. This is intended to be used with the KVS WebRTC C SDK running as the master and the JS SDK as the viewer. The master sends peer, ice-agent, signaling and data-channel metrics to the viewer which are plotted ~ 20 seconds after the viewer is started. Since the timeline plot is intended to understand the time-to-first-frame, the sample web page needs to be refreshed and the master needs to be restarted if a new / updated plot is needed. While using the SDK in this mode, it is expected that all datachannel messages are JSON messages. This feature is only meant to be used for a single viewer at a time.

Category	Metric	Calculation	Description
viewer-wait	viewer-waiting-for-master	Calculated as the time between the signaling client going to open state and the time to start the master	Time duration the viewer was waiting for the master to start (time to start the SDK after the viewer signaling channel was connected)
viewer-signaling	signaling-viewer	Calculated as the time between creation of the KinesisVideoClient and the signaling client going to open state	Time taken to establish a signaling connection on the viewer-side
viewer-signaling	setup-media-player-viewer	Calculated as the time taken for getUserMedia call	Time taken to setup a media player on the viewer-side by seeking permissions for mic / camera (if needed), fetch tracks from the same and add them to the peer connection peer.
viewer-signaling	signaling-viewer-describe-channel	Calculated as the time taken for the describeSignalingChannel call	Time taken for the API call to describeSignalingChannel on the viewer
viewer-signaling	signaling-viewer-describe-media-storage-config	Calculated as the time taken for the describeMediaStorageConfiguration call	Time taken for the API call to describeSignalingChannel on the viewer
viewer-signaling	signaling-viewer-get-signaling-channel-endpoint	Calculated as the time taken for the getSignalingChannelEndpoint	Time taken for the API call to getSignalingChannelEndpoint on the viewer
viewer-signaling	signaling-viewer-get-ice-server-config	Calculated as the time taken for the getIceServerConfig	Time taken for the API call to getIceServerConfig on the viewer
viewer-signaling	sign-connect-as-viewer	Calculated as the time to sign the websocket url to connect to Signaling via connectAsViewer	Time taken to sign the websocket url
viewer-signaling	signaling-connect-as-viewer	Calculated as the time the Websocket object is created with the signed url, to it reaching the open state	Time taken to establish a connection to Signaling via connectAsViewer
viewer-sdp-exchange	sdp-exchange-viewer	Calculated as the time between setting the local description and the reception of an answer from the master	Time taken to send an offer and receive a response
viewer-ice	ice-gathering-viewer	Calculated as the time between viewer.peerConnection.iceGatheringState going from gathering to complete	Time taken to gather all ice candidates on the viewer
viewer-peer-connection	pc-establishment-viewer	Calculated as the time between viewer.peerConnection.connectionState going from new to connected	Time taken to establish the peer connection on the viewer
viewer-datachannel	datachannel-viewer	Calculated as the time between timestamp1 (time at which the viewer sends a message) and timestamp3 (time at which the viewer receives a response from the master)	Time taken to send a message to the master and receive a response back
viewer-ttff	ttff-after-pc-viewer	Calculated as time between viewer.peerConnection.connectionState in connected state and viewer.remoteView.addEventListener goes to loadeddata	Time to first frame after the viewer's peer connection has been established
master-wait	master-waiting-for-viewer	Calculated as the time between signaling state going to connected and the button to start the viewer	Time duration the master was waiting for the viewer to start (time to click the button after the master signaling channel was connected)
master-signaling	signaling-master	Calculated as the time for the entire signaling process on the master	Time taken to establish a signaling connection on the master-side
master-signaling	signaling-master-describe-channel	Calculated as the time taken for the describeSignalingChannel call	Time taken for the API call to desribeSignalingChannel on the master
master-signaling	signaling-master-get-signaling-channel-endpoint	Calculated as the time taken for the getSignalingChannelEndpoint	Time taken for the API call to getSignalingChannelEndpoint on the master
master-signaling	signaling-master-get-ice-server-config	Calculated as the time taken for the getIceServerConfig	Time taken for the API call to getIceServerConfig on the master
master-signaling	signaling-master-get-token	Calculated as the time taken for getCredentialsFn for authentication	Time taken for the getToken call on the master
master-signaling	signaling-master-create-channel	Calculated as the time taken for the createSignalingChannel	Time taken createChannel API call on the master
master-signaling	signaling-master-connect	Calculated as the time taken for the connectAsMaster	Time taken for the signaling connect on the master
master-sdp-exchange	sdp-exchange-master	Calculated as the time between the master receiving an offer, processing it and sending an answer to the viewer	Time taken to respond to an offer from the viewer with an answer
master-ice	ice-gathering-master	Calculated as the time from starting the gathering to the time its scheduling was stopped	Time taken to gather all ice candidates on the master
master-peer-connection	pc-establishment-master	Calculated as the time taken to go to RTC_PEER_CONNECTION_STATE_CONNECTED	Time taken to establish the peer connection on the master
master-datachannel	datachannel-master	Calculated as the time between timestamp2 (time at which the master sends a message) and timestamp4 (time at which the master receives a response from the master)	Time taken to send a message to the viewer and receive a response back
master-ttff	ttff-after-pc-master	Calculated as the time between metrics.master.peerConnection.endTime and viewer.remoteView.addEventListener reaches loadeddata	Time to first frame after the master's peer connection has been established
ttff	ttff	Calculated as the time taken between the viewer button is clicked and viewer.remoteView.addEventListener reaches loadeddata	Time to first frame since the viewer button was clicked

Compatibility

The SDK is supported in the following browsers / environments:

Chrome	Edge	Firefox	IE	Safari	Android Webview	Android Chrome	iOS Safari	NodeJS
52	12	36	No	11	53	52	11	8

To increase WebRTC API compatibility between different browsers, it's highly recommended to use adapter.js.

Following is a quote from adapter.js docs:

adapter.js is a shim to insulate apps from spec changes and prefix differences in WebRTC. The prefix differences are mostly gone these days but differences in behaviour between browsers remain.

Managing Credentials

The SignalingClient requires a SigV4 signed URL in order to make requests to the KVS signaling service backend. The client can either be provided with AWS credentials (and then it will use those to sign requests) or it can be provided with a custom RequestSigner that can perform the request signing.

There are several mechanisms that are recommended for managing AWS credentials in a web client, such as using Cognito or Federated Identities, that are explained in the AWS SDK for JS documentation.

Alternatively, if you do not want any AWS credentials in the web client, you can provide a request signer that makes a call to your own backend that uses AWS credentials to create a signed request for the KVS WebRTC Signaling Service. With a NodeJS based backend, you can create signed requests using the SigV4RequestSigner class. Note that you will also have to get other data, such as the ICE server config, on the backend and send that to the client.

IAM Permissions

Regardless of the mechanism used to manage the credentials, the credentials will need to have permissions to perform KVS operations. The following is an example policy for a viewer of a particular channel:

1{
2  "Version": "2012-10-17",
3  "Statement": [
4    {
5      "Sid": "kvsViewerPolicy",
6      "Action": [
7        "kinesisvideo:ConnectAsViewer", // Use "kinesisvideo:ConnectAsMaster" for master policy instead.
8        "kinesisvideo:DescribeSignalingChannel",
9        "kinesisvideo:GetIceServerConfig",
10        "kinesisvideo:GetSignalingChannelEndpoint"
11      ],
12      "Effect": "Allow",
13      "Resource": "arn:aws:kinesisvideo:<region>:<account_ID>:channel/<channelName>/<creationTime>"
14    }
15  ]
16}

See KVS WebRTC Access Control Documentation for more information.

Development

Running WebRTC Test Page Locally

The SDK and test page can be edited and run locally by following these instructions:

NodeJS version 16+ is required.

1. Run npm install to download dependencies.
2. Run npm run develop to run the webserver.
3. Open the WebRTC test page at http://localhost:3001

You will need to provide an AWS region, AWS credentials, and a Channel Name.

The source code for the test page is in the examples directory.

For advanced debugging, use the WebRTC Internals tool your browser provides.

License

This project is licensed under the Apache-2.0 License. See LICENSE.txt and NOTICE.txt for more information.