23
Multiplayer games on networks potential and tradeoffs

Multiplayer games on networks potential and tradeoffs

Embed Size (px)

Citation preview

Page 1: Multiplayer games on networks potential and tradeoffs

Multiplayer games on networks

potential and tradeoffs

Page 2: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Brief history of real-time networking

approximately...1980’s (US) military simulations1990’s distributive / collaborative virtual

environments2000’s multiplayer gamesbut

1980 1990 2000

MUD Amaze DoomUltima Online

Battle.net

Page 3: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Analyzing constraints on network games*

design constraints at three levels physical: cabling and hardware -

bandwidth, latency, processing logical: architectures for communication

(e.g., peer-to-peer, client/server), data and control (e.g., centralized, distributed, replicated)

application: interpretation (data meaning), integrity control (legal state), transferring (computation shared among nodes)

*Smed, Kaukoranta & Hakonen, 2003

Page 4: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Resource management – physical and logical influence on experience

resource requirements in distributed application: f(info sent, info received, processing rate)

IPE Information Principle Equation: Singal &Zyda, 1999

M: number of messagesH: average number of destination nodesB: bandwidth required for message to one destinationT: delayP: number of processor cycles to receive, process a

message

PTBHMresources

Page 5: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

IPE

product is a measure of“quality of experience”

reducing a factor reduces resource requirements but also reduces experience

factors can be traded off against each other

Page 6: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

IPE exampleM: number of

messagesH: average number of

destination nodesB: bandwidth required

for message to one destination

T: delayP: number of

processor cycles to receive, process a message

PTBHMresources

packet aggregation M decreased B increased P increased T increaseddesign question – is

tradeoff valuable?

Page 7: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Quality of experiencein networked game

consistency: similarity of view to actual data in the nodes of the network – “absolute consistency” is perfect data view match among nodes

responsiveness: – delay for an update event to be applied to game data

in database applications – consistency takes precedence over responsiveness

in game applications – responsiveness is more important and consistency may be compromised

Page 8: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Data and control architectures

centralized distributed replicated

process relay data store network

Page 9: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Consistency, responsivenessand architecture

high consistency processes on remote nodes tightly coupled

high responsiveness processes loosely coupled

Page 10: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Relay

local input

global output

global input

local output

networknode

2 – way relay assures consistency in combination with distributed or centralized architecture BUT responsiveness is vulnerable to network

Page 11: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Relay

short-circuit relay assures high responsiveness but local data can become inconsistent with other nodes

local input

global output

global input

local output

networknode

Page 12: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Resolving inconsistencies

games require responsiveness inconsistencies will occur mechanism for resolving

inconsistencies must be provided

Page 13: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Application level influence on experience

scalability – adapting to resource changes

persistence – effect of players leaving and entering

collaboration – consistency and sharing

Page 14: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Scalability

game server issues allocation of non-player computation

among nodes parallelism & concurrent computation model each node burdens network but increases

computation power ( H up, P down)

Page 15: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Persistence

relationship of node to application initializing state of node when player joins on-going consistency maintaining state when node leaves game abrupt disruption – reallocation of

responsibilities

Page 16: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Collaboration

teams acting together communication among team members

prioritized messages / dedicated channel ‘interest management’ based on

application-based closeness or ‘collaboration distance’

Page 17: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Reducing resource requirements

Architecture level packet compression

internal – data compression within packet suits UDP

external – data compression bsed on assumed receiver state

packet aggregation merging removes some headers but adds

delay to data waiting for packetizing timeout or quorum-based aggregation tradeoffs

Page 18: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Reducing resource requirements

Application level interest management

premise – most packets only of interest to minority of nodes

transmit only to those, not all nodes nodes ‘express interest’ in subset of

information M down, B down, P up

Page 19: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Interest management

area of interest – ‘aura’ application based – correlates to sensory

capabilities of node system (typically character)

aura = focus(perception) + nimbus(perceptibility)

when focus intersects nimbus, nodes communicate

Page 20: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Nimbus and Focus Smed et al,

2003

Page 21: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Interest management

area of interest – ‘aura’ ‘subscription manager’ server manages

state, interest info for each node subscription manager receives messages

about state and retransmits to interested nodes

nodes submit changes of state and interest to the subscription manager

Page 22: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

Improving responsiveness

e.g., with less frequent packets due to aggregation

game state extrapolation techniques to continue response without input ‘dead reckoning’ based on predicting

state changes and correcting when info arrives

Page 23: Multiplayer games on networks potential and tradeoffs

COSC 4126 network games

dead reckoning example Smed et al, 2003based on recent

positions, compute velocity, maybe acceleration and extrapolate

when update arrives, plot ‘smooth’ convergence