Multi-user

[biyue111]

• need to observe more than one timeslot (change state s to a list )
• add reward function in environment(channel utility, fairness, less channel change), which can also make the code easier to read
• because there are multiple users, env.step need to accept a row of action a and return a column of next state s_
• To calculate reward, ChannelEnv class need to store the actions of all user in T timeslots.
• Add jammer's histogram

[biyue111]

pre-work to make the code simpler

• set action to 1 -- channel_cnt
• eliminate all terminal, is_terminal etc, which are totally useless in our case
• use no more key="%d-%d"%(state, action) if possible
• use functions isChannelBlocked and getChannelNumber to convert state to channel's availability and channel number