Management of Broadband Media Assets on Wide Area Networks

1
Management of Broadband Media Assets on Wide Area
Networks

• Lars-Olof Burchard

2
Motivation TV Anytime Services

• Web Services anything, anywhere, anytime
• Vision Access to broadband media content
  delivered by TV channels at any time
• Limiting factor Performance of networks will
  never be enough to support interactive access to
  masses TV content (in MPEG quality)
• Approach Install (caching, mirror) server
  systems close to the user
• Advantage Fast and cheap access for local users
• Disadvantage large storage costs for audio/video
  content
• Idea of TV Anytime Store large quantities of
  digital TV content on local servers
• Content distribution using classical networks
  satellite, cable

3
Implementation of TV Anytime

• Options for server implementation of TV Anytime
  services
• Local server, e.g. within the living room of a
  user
• Distributed server system public server
  installed within ADSL, CATV central office and
  direct line from server to local user (not
  congested)
• Advantages local server
• Cheap access to stored content items (media
  assets)
• Easy content management (access management,
  security, billing, user data, ... )
• Advantages distributed servers
• Popular content items can be shared by many user
  (minimisation of overall storage)
• Large servers allow the provision of more content
  
• Our solution
• Hierarchy of small local servers (home servers)
  and larger networked servers

4
Hierarchical VoD servers for TV Anytime services
Broadband Network
LAN
PC
PC
PC
PC
LAN
Cable Network
PC
PC
TV
TV
PC
TV
PC
TV
5
Issues for hierarchical TV Anytime server networks

• Technical issues building up coherent network of
  CATV, ADSL, ATM, ... lines
• User management, access management, legal issues,
  ....
• Decision for recording / deletion of media asset
  
• Decisions about recording / deletion have to be
  made
• Build virtual large server that collects all
  clients that are connected to the sub-tree
• Studied here Decision of placement of media
  asset on server hierarchy (using different
  qualities, i.e. bit rates)
• Trade-off in placement of media assets onto
  hierarchical server network between available
  bandwidth/storage space and bit rate
• Solution Simulated annealing algorithm to solve
  combinatorial optimisation problem

6
Solution strategy for combinatorial optimization
problem

• Goal maximisation of QoS (i.e. bit rate of
  requested assets)
• Simulated Annealing as classical (problem
  independent) solution method
• Proven strategy for combinatorial optimisation
  problems
• Static scenario
• compute assignment of a fixed set of assets
  (based on known user behaviour using profiling,
  subscriptions)
• Dynamic scenario additional assets are recorded,
  others deleted during run time
• reduction to static problem

7
Results

• Benchmarks trees with varying number of nodes,
  various amounts of assets
• Benchmark results comparison to upper bound
• BIN_7 4.31 , BIN_15 4.80 , BIN_32 5.25
  , BIN_64 7.88
• TER_4 4.36 , TER_14 4.22 , TER_40 9.13
• QUAD_5 4.69 , QUAD_21 6.23 , QUAD_85
  9.44
• within 4 to 10 to upper bound of SA heuristic
  (smaller difference to optimal solution)