Sliding Window

• Requires propablity optimization
• Solution should be re-evaluated for each $T_W$ sliding across $T_{max}$
  • Re-evaluation should use efficient rerun of bellman ford
  • Update distribution of current NS
  • Set future nodes of current NS weight to inf
  • keep other nodes same weight as before
  • Maybe think about better optimizations ?
• Instead of sliding window, maybe only run once at $t_1$ after new connection
  • This way we assume that our ofdma allocation is constant in time
  • Even more complex would be to have ofdma last for some time $T_O$ and evaluate if resources change