Verry Nice HighFive

1
Verry Nice! High-Five!

• Yoojin Kwak
• Michael Preysman
• Ruben Quintero
• Russell Savage

2
Test Cases

• Concept
• Portable communication device for use within
  industrial, collaborative work environment
• Status
• Back end mostly finished, GUI in the works.
• Gumstixproblems with ALSA and GNU Classpath
• Test Cases
• Boot Gumstix, connect on boot. Connect multiple
  Gumstix to the server.
• Send/receive audio from the base station to the
  Gumstix.
• Server acceps multiple users and can move them
  between groups.
• Server talks only to specific groups and/or users.

3
Performance Tests

• Latency (primary)
• Why Ability to have a normal conversation.
• How Stopwatch/Basic Usability. Test Latency vs.
  Traffic.
• Network Traffic
• Why How many users can we support?
• How Increase number of users speaking and
  connected
• Packet Loss
• Why Dont want clipped, unrecognizable speech.
• How JSpeex handles packet loss.
• Seamlessness
• Why Users should be able to move between nodes
  (mobility).
• How Piggy backing on existing networks.

4
Testing Process

• Network Traffic
• Brute force increase users until system fails.
• Automated increase packets sent (not voice,
  just random info).
• Latency
• Ping from the Gumstix.
• Stopwatch/General Usability.
• Also increase users, see how latencyis affected.

5
Difficulties

• Latency will be most difficult because it is
  affected by a variety of things (traffic,
  processing, packet transfer, etc).
• Pinging does not time from input at the
  microphone to output from the speakers
  (compression, processing, human reaction time).
• Difficult to find the source of certain errors
  (for instance, sound Classpath, JamVM, ALSA,
  OSS, etc.)
• Gumstix have been relatively difficult to work
  with.

6
Question Time!