PowerPoint Presentation

Simplify video streaming for

millions of smart home cameras

Girish Sood

A N T 3 1 8

Senior Product Manager

Amazon Kinesis Video Streams

Amazon Web Services

Frederik Delacourt

Sr. Director

Technology & Services

Wyze Labs

Tao Yuan

Director

Software Development

Wyze Labs

Agenda

• Camera devices in smart homes

• Overview of Amazon Kinesis Video Streams

• Launching – Support for WebRTC in Kinesis Video Streams

• Demo

• Wyze Labs

Smart Home: Use Cases

Home Automation

Lighting systems

Media & entertainment

White goods & appliances

Small appliances

Home assistants

Soundbars

TVs

Vacuum robots

Home Security & Monitoring

Door locks

Thermostats

Security cameras

Water leak detectors

Video Doorbells

Home Networking

Wi-Fi routers & modems Cable set-top boxes

Media Streaming for Camera Devices: Basics

Store

Securely and durably store

the media

Retrieve

Real-time and on-

demand retrieval for

playback and processing

Stream

Reliably stream audio

and video data

Media Streaming for Camera Devices: Advanced

Connect

Seamless integration

with smart home

Understand

Machine learning for

intelligent insights

Media Streaming for Camera Devices

• Securely and reliably ingest media from the device

• Durably store, encrypt, and index media

• Playback media—live and on-demand

• Process media for ML, and other real-time and batch-mode processing

Complexity

• Support multiple media streaming technologies

• Support multiple devices and development environments

• Build and operate infrastructure for secure and reliable streaming

• Elastically scale to support millions of devices

• Build engineering teams with video streaming expertise

Kinesis Video Streams: Fully Managed Streaming

• Securely ingest media from

millions of devices

• Store and time-index media,

search for media via simple APIs

• Consume and Process media for

• Live and on-demand playback

• Real-time and batch-mode processing

Amazon

Rekognition

Video

Amazon

SageMaker

HLS and

DASH

Playback

Custom

Video

Processing

Third-Party

Partners

Kinesis Video

Streams

Kinesis Video Streams: Ingesting Media

• Stream directly from

(supported) cameras

• Use a proxy/gateway that

connects with device on the

same network

• Both approaches use the Kinesis

Video Streams producer SDKs

Gateway

Kinesis Video Streams: Producer SDKs

Flexible SDKs for integration with on-device hardware media pipelines

Platform SDKs

Low-level SDKs

Docker images

Platform independent layer (C)

OO Wrapper Layer (C++ and Java)

Android

Win

Linux

MacOS Raspian

Docker Image

Gstreamer

plugin

Gstreamer

plugin

Gstreamer

plugin

Gstreamer

plugin

Kinesis Video Streams: Storing and Indexing Media

• Video streams: Managed

resource for ingesting media

with time-indexed storage

• Easily monitor and audit usage,

apply access control and

authentication policies

• Easy APIs to apply retention

policies, and retrieve live and

recorded media

• Encrypted in-transit and at-rest

Store Encrypt Index

Kinesis Video Streams: Processing for ML

• Process real-time or batch-mode

media frames with AWS Deep

Learning AMIs

• Amazon Kinesis Video Inference

Template for Amazon

SageMaker

• Amazon Rekognition video for

fully managed machine learning

based processing

Amazon

Rekognition

Amazon

SageMaker

AWS Deep

Learning AMIs

Kinesis Video Streams: Processing for Playback

• Live and on-demand playback

with HTTP Live Streaming (HLS)

and Dynamic Adaptive

Streaming over HTTP (DASH)

• Support for multiple audio and

video encoding formats

Mobile

and Web

Players

Camera

Devices

Kinesis Video Streams: WebRTC

Stream live media with ultra-low latency and enable two-way interactivity for

millions of camera devices

Standards

Compliant

Exchange audio, video, and data

b/w devices, mobile, and web

apps for real-time, two-way

interactivity

Fully

Managed

Fully managed WebRTC

signaling, TURN, and STUN

services with easy to use SDKs

Real-time, Two-way

Interactivity

Compliant with web and

mobile platforms for easy

plug-in free playback

Low Latency Live

Media Streaming

Peer-to-peer audio and video

live streaming with sub-1 second

latency for playback

WebRTC: Technology Overview

Connectivity

Establish peer-to-peer

connectivity

Encryption

End-to-end

encryption

Media Delivery

Low latency exchange

media and arbitrary data

Signaling

Exchange connection

metadata

WebRTC is not “just” a media streaming protocol. It is an open standard for real-

time communication with technology specifications for:

WebRTC Live Streaming: Example Workflow

Signaling

TURN

STUN

Kinesis Video Streams: WebRTC

• Fully managed services for WebRTC Signaling, TURN, and STUN

• Open source embedded SDKs for IoT devices

• Open source client SDKs for mobile and web applications

Kinesis Video Streams WebRTC: Services

• Signaling

• Signaling channels—AWS resource name (ARN) with easy to use APIs

• Signaling message transport is backed by secure WebSockets

• Purpose built for smart home camera use cases

• TURN

• Support media relay when P2P streaming is unviable

• STUN

• Enables P2P connectivity for devices behind NAT

Kinesis Video Streams WebRTC: SDKs

Embedded SDKs

• Open source platform-independent

SDKs for IoT camera devices

• Supports Trickle ICE for faster

connectivity

• Media and data

• H264 and VP8 for video

• Opus and G711 for audio

• Standards-compliant

implementation

Client SDKs

• Open-source client SDKs that

seamlessly work with WebRTC

compliant browser and mobile

platforms

• Signaling clients

• JavaScript, Android, and iOS

Kinesis Video Streams WebRTC: Alexa Integration

Kinesis Video Streams WebRTC

Alexa smart home skill

Video Doorbell Alexa Device

Two-way audio and video

Wyze Labs

Frederik Delacourt

Sr Director Technology & Services

Tao Yuan

Director of Software Development

Who is Wyze Labs?

• Seattle startup

• Create a convenient and

enjoyable IoT lifestyle

• By making great technology

accessible to everyone

Growth

2.5M Customers

3M+ Devices

Activated

Features & Services

C

K

Business Best Practices

Focus on differentiating feature set

Push all commodity features to suppliers

Features & Services

C

K

C

Wyze Event Video Architecture: Before

Requirements

• 12s completed videos only

• No live video need

Wyze Cloud work

• Amazon S3 access management

• DevOps (load balancing, metrics)

New Requirements

• Long videos for up to 5 minutes

• Can view live video

Need a new solution!

x

Wyze Event Video without Kinesis Video Streams

Wyze Cloud new work:

• Video stream storage

management

• Video stream security

management

• More DevOps work

Firmware new work:

• Slice video/audio into

fragments

• Fragments upload

management (buffer, sync,

retry, etc.)

Wyze Event Video with Kinesis Video Streams

Benefits

• Ingest/digest is simpler and

safer

• No stream management

hassle for Wyze

• Don’t need to worry about

scaling problem. No DevOps ☺

• A suitable solution for long

videos and future scenarios

Wyze Architecture: Potential Impact of WebRTC

Current Architecture Features

• Live streaming:

protocol

• Event video: Amazon S3

• Alexa integration: RTMP + RTSP

• Google Home integration: RTMP +

MPEG-DASH

• Two-way Audio: N/A

• Local Connection: N/A

Multiple players in mobile apps

Multiple IoT registrations

Wyze Architecture: Potential Impact of WebRTC

Current Architecture Features

• Live streaming: WebRTC

• Event video: WebRTC

• Alexa integration: WebRTC

• Google Home integration: WebRTC

• Two-way Audio: WebRTC

• Local Connection: Possible

Single player in mobile apps

Single IoT registration

• Live streaming:

Protocol

• Event video: AWS S3

• Alexa integration: RTMP + RTSP

• Google Home integration: RTMP +

MPEG-DASH

• Two-way Audio: N/A

• Local Connection: N/A

Multiple players in mobile apps

Multiple IoT registrations

Potential Architecture

Conclusion

Large scale secure

media ingestion and

retention

Encrypted, durable, and

time indexed media storage

Real-time and on-

demand playback

and processing

HLS, MPEG-DASH, and

integration with AWS ML

services

Low latency live

streaming and two-way

interactivity

WebRTC audio, video,

and data

Amazon Kinesis Video Streams is a fully managed service that elastically scales to

support the media streaming needs of millions of camera devices

Learn big data with AWS Training and Certification

Visit aws.amazon.com/training/paths-specialty/

New free digital course, Data Analytics Fundamentals, introduces

Amazon S3, Amazon Kinesis, Amazon EMR,

AWS Glue, and Amazon Redshift

Validate expertise with the AWS Certified Big Data - Specialty exam or

the new AWS Certified Data Analytics - Specialty beta exam

Resources created by the experts at AWS to help you build and validate data analytics skills

Classroom offerings, including Big Data on AWS, feature

AWS expert instructors and hands-on labs

Thank you!

Garish Sood

garisood@amazon.com