SMN 1.0 Smart Media Network Auburn University COMP7970 Richard Chapman 19 Sept 2002.

SMN 1.0Smart Media Network

Auburn University

COMP7970

Richard Chapman

19 Sept 2002

Home audio/video

Back side view...

And…

How do real users cope?

The problems

• Undocumented configuration and interconnection information

• Difficult to modify or reconfigure: snarl of wire

• “Secret” knowledge required to operate

• Sometimes just cannot be configured

Motivations for a solution

• MP3, streaming media, fast Ethernet

• Jini, CORBA, OSGi, UPnP

• Universal remote controls based on PDA, Mobile phone, web

• Utter indifference of consumer electronics manufacturers

Calls for a solution

• “A Call for the Home Media Network”, Bell and Gimmell, Communications of the ACM, July 2002, pp. 71-75

• Donald Norman, “The Perils of Home Theater”, IEEE Computer, June 2002.

Requirements

• Good audio and video quality

• Easy to operate

• Easy to configure, udpate, modify

• Easy to develop for

A bad solution

Other goals

• Open source• Standards based• Vendor, OS

language neutral• Keep orthogonal

issues orthogonal• Pay attention to the

user interface

• No global state to maintain

• Capable of redundancy, fault tolerance

• Support multiple UI’s• No preferences or

settings

User stories

• “I want to play CD’s, video or listen to the radio”

• “I want to record my soaps”

• I can listen to music or watch TV or movies”

What can we deduce

• Users want to ignore connectivity

• Users don’t want to set up the system

• Content is what matters

A developer story

An architecture

Device categories

• Server -- multiple okay, no direct user access• Controller (phone, PDA, web, custom HW) -- multiple

controllers okay• Player -- originates content stream (audio, video

players, tuner, Internet radio)• Recorders (PVR, MP3 ripper, etc)• Presenters(amp/spkrs, video monitor)• Processors (e.g. EQ, reverb)

What is good

• The snarl of wires is gone

• The configuration information is no longer secret

• Control and data can share one wire

What might be bad?

• Audio and video quality?

• Use RTP/UDP

• Buffering is okay

• Compare Gibson Guitar Corp MaGiC, Peak Audio (synchronous protocols for audio over Ethernet, but not TCP/IP)

• We’re going to try TCP/IP

The orthogonal issues

• Service registration, lookup, discovery

• Control messages to devices

• Streaming protocol

• User interface

Service registration, leasing, lookup, query

• Many options exist: OSGi, UPnP, Jini, CORBA, SOAP

• We have experience with Jini (smart badges, lego demo, eClassroom)

• Want language and OS independence

• Want something simple

Problems

• CORBA would require “brittle” interface wrappers (IDL) for everything

• Jini has more flexibility, but still requires Java on all devices

• UPnP -- is Microsoft ready for a university development effort?

SMN 1.0

• Use LDAP for registry• Add a “keepalive daemon” to handle

leasing• Use SOAP for control messages• Use XML to store description of device

capabilities, control message formats, in the server

• Use variety of streaming protocols

Why LDAP?

• Solves the registry problem

• Simple to implement

• Language, vendor, OS independent

• Lots of languages have LDAP libraries

• OpenLDAP

How much should the controller know about a device?

• Nothing: just present the controls to the user in some “nice way” on the interface

• Lots: this way you get as much basic functionality as you can

• “The universal remote is always missing a few buttons you really need”

LDAP Schema

User Interface Issues

• Make the common tasks simple• One way of doing things• No “set preferences” in the interface• Support multiple interface

implementations• No configuration knowledge required• Get away from WIMP and remote

controls as interfaces

One knob, one button, one navigator

button

trackball

rotating ring(volume)

Selection screen

PalmOS remote

PalmOS remote

What we will do

• Define the LDAP schema, SOAP messages, XML format for the server

• Implement a basic player, presenter

• Implement a couple of controllers

• Probably audio only

• User study

What else is there to do?

• OSGi gateway to the WAN (mobile phone controller, mobile phone presenter, multi-LAN network)

• Other controller implementations

• Alternative service discovery, registration system