1. Paper Review: Latency Evaluation of Networking Mechanisms for Game Traffic Jin, Da-Jhong

2. Motivation  Most interactive games need response time between 100 ~ 1000 ms depending on game genre.  Challenging system requirement of interactive games: low latency for all users.  In this paper, they evaluate different techniques for delivering packets in timely manner.

3. Game Trace Analysis: Game Stream Characteristics  Game trace analysis: –Small packet size –High interarrival time between packets.

4. Experiment Setting  Because Small pkt size, high interarrival time between pkts, so they test under these condition: –Small pkt: 100 bytes –Pkt sent at low rate: interarrival time 50, 100, 200ms –RTT: 50, 100, 200ms –Loss rate: 0.1%, 0.5%, 2.5%  Emulator: netem  Test latency vs RTT under fixed IT and loss.  Test latency vs IT under fixed RTT and loss.  Test latency vs lost rate under fixed IT and RTT.

5. Network Mechanisms  ENet (Developed for FPS Cube engine)  Modified ENet (Exponential backoff removed)  UDT (UDP-based Data Transfer Protocol)  TCP new reno (variation of TCP)  TCP bic (variation of TCP, binary increase congestion control)  Modified TCP (Exponential backoff removed)  TCP with RDB (variation of TCP, redundant data bundling)  SCTP  Modified SCTP (Enable bundling and fast retransmission)

6. TCP new reno  The main difference between each TCP variants is congestion control.  TCP new reno: 1.When 3 dupACKs are received, set ssthresh and cwnd are equal to ½ cwnd. 2.When another dupACK is received, cwnd + 1 3.When a new pkt’s ACK is received, recover cwnd to the number in step 1.

7. TCP BIC  TCP bic: binary increase congestion control, performs a fair share of the bandwidth faster than new reno. –Binary search: the value of cwnd jump to the middle point of wmax and wmin.

8. Enhancement for Thin Stream  Modified TCP: exponential backoff removed to avoid the extreme latencies that occurs when consecutive packets are lost.  TCP with RDB: redundant data bundling, enable to bundle unacknowledged data if there is room in packet. Thus, loss packet may be recovered when next packet is received without retransmission.

9. ENet  Not a conjunction of TCP & UDP, but a single, uniform protocol layered over UDP. –Sequencing  Assign to each sent packet a sequence number that is incremented as packets are sent.  Guarantee that no packet with a higher sequence number will be delivered before a packet with a lower sequence number.  Stall delivery of the higher sequence number packets until its predecessors have arrived. –Reliability  Optional reliability of packet delivery by ensuring the foreign host acknowledges receipt of all reliable packets.  Modified ENet: exponential backoff removed.

10. UDT  UDT: UDP-based Data Transfer Protocol –UDT is designed to co-exist with TCP (differ from ENet)

11. SCTP  SCTP –Fast retransmission and timeout mechanisms as in TCP  Modified SCTP: exponential backoff removed

12. Latency vs. RTT

13. Latency vs. Interarrival Time

14. Latency vs. Loss Rate

15. Explanation  Average results are similar, except SCTP has higher latency.  Worst case results: –It is due to consecutive retransmissions of the same packet. –Modified protocol is better than not modified one.  Removing exponential backoff can improve latency. –TCP with RDB behave better because send multiple copies of a data element by bundling unacknowledged data in a packet.

16. Bandwidth

17. Price of Retransmission Latency  Traditional TCP needs less bandwidth.  UDT and TCP with RDB need higher bandwidth. –UDT always tried to use all the estimated bandwidth. When RTT is low, UDT will resend a packet multiple times to fill the pipe. –TCP with RDB bundles previous packets as long as there are unACK data.  Reduce the retransmission latency increases bandwidth requirement.

18.  Strength: –Widely measure existing protocols and middleware which can be used to run games.  Weakness: –There is only one packet size, 100 bytes.

19. Thanks for Listening!