Incentives - Peer-to-Peer Computing: Applications, Architecture, Protocols, and Challenges - page 119

TABLE 5.6: Symbols.

Symbol Definition

r rate of the media content

t duration of the media content

n number of stripes

P RX
s

receive power of the server interface

R s data rate of the server interface (downlink)

P RX
p

receive power of the peer interface

R p data rate of the peer interface (symmetric)

E RX
s

(i) energy cost of receiving i stripes through the server inter-

face (i≤n)

E RX
p

(i) energy cost of receiving i stripes through the peer interface

(i≤n)

E T X
p

(i) energy cost of transmitting i stripes through the peer in-

terface (i≤n)

α x client x's type

α 1 (i) threshold value for a client to receive i stripes from the

server

α M (i) threshold value for a master to receive i stripes, i≤n

α S (i) threshold value for a slave to receive i stripes, i≤n

α ′ r (i)

threshold value for a coordinator with (r−1) helpers to
receive i stripes

α r (i) threshold value for a helper to receive i stripes

constraints. Since clients are autonomous entities, we should also consider their
incentives for collaboration. Specifically, each client is modeled as a selfish but
rational entity whose degree of selfishness is characterized by its type. Here,
client x would only collaborate with other clients provided that the number of
stripes x obtained from collaboration is no smaller than s 1 , where s 1 is given
by:

s 1 = argmax ∀i≤n {α≤α 1 (i)} (5.50)
This means that each client is only interested in improving its own stream-
ing performance. We note that the type of a client quantifies its selfishness.
For example, when α x = 0.8, client x would collaborate with other clients for
media streaming if it can receive more stripes from collaboration; otherwise,
x would rather stream the media content from the server independently. It is
then interesting to study the feasibility of collaboration among these selfish
clients.
As in all practical P2P data sharing systems (e.g., BitTorrent), there are
always some enthusiastic participants that would be willing to spend more re-
sources in return of a larger participating population. Such enthusiastic clients
can be modeled as masters which have a larger value of α, as detailed below.