Transcript Online Game network traffic optimization

ONLINE GAME NETWORK
TRAFFIC OPTIMIZATION
Jaewoo kim
Youngho yi
Minsik cho
INTRODUCTION
• Optimizing Online Game Traffic
• Interested in Improving poor network performance
• Contributing open source project
• Game: Speed Dreams 2
•
•
•
•
Open source motorsport simulation and racing game
Released under the GNU General Public License (GPL)
C++, OpenGL
ENet: network Library
CURRENT ISSUES
• Speed Dreams 2 provides very poor network game
experience
• Lags and Disconnection occurs frequently
• Speed Dreams 2 broadcasts packets as much as it can
• One instance is responsible for all network connection
• Speed Dreams 2 always uses the same port (28501) for p2p
• This prevents us from running many instances on the same
computer for testing purpose
• In addition, if other process has already used 28501 port, then the
user cannot play network mode
• Some of bug needed to be fixed in the overall game code
OPTIMIZATION
• Modifying Architecture for stable network environment
• Currently host client acts as server and client
 Centralized P2P in order to support massive users
• Game State Management
• Traffic occurs between all players even farthest players.
• Define “Interest area” Remove unnecessary traffic
by applying interest management
• Multicast only within the same area, as player moves,
the interest area is constantly updated
• Sending packets too often between clients
• Will decrease number of packets per second
• Also reducing amount of packets by compressing packets
EVALUATION
• Ultimate goal is proving a quality gaming experience
with optimization techniques
• Comparison through recorded videos between current
and improved version of Speed Dreams 2
• Metrics to measure the performance
•
•
•
•
Delay Variation (Jitter)
Ping (Latency)
Packet Loss Rate
Bandwidth
DEVELOPMENT/EXPERIMENT
• Used open source project, Speed Dreams 2
• Bugs fixes
• Solved assigning port number problem to instances on a same
client -> port number is assigned dynamically
• Defined Metrics to evaluate
• Applied techniques:
• Interest Management
• Data compression / Decreasing number of packets
• Will test network performance using metrics
• Will adjust parameters and slightly modify network architecture
Thank you