Adaptive Reward Allocation for Participatory Sensing

Algorithm 1

Reward computation.

// Construct a Linear Regression model from the supply curve.

Create Supply Curve for from Historical Dataset

Prediction_Model = Linear Regression Model for

Supply Curve

// Predict the responses for different reward levels.

= 0..

] = predict(Prediction_Model, )

// Construct the queuing state variables for each reward.

foreach R

= -

R, , (t)

end foreach

// Compute the constant used for Lyapunov Optimization.

= 0.5 ((t-1) - (t))²

// Map the data utility weighting U to the value of V

// Evaluate each reward using Lyapunov Optimization.

foreach R, , (t)

// Compute the Budget used by this reward.

B = R

//Check that the budget consumption

// does not exceed the set maximum.

if B > then

 break

// Carry out the Lyapunov Optimization

// computation.

L = 1/2 (t)²

// Compute the one slot conditional Lyapunov drift.

= L - (R)

// Evaluate the drift plus penalty expression.

=

= + (V B)

+ Z(t)((t-1) - (t)))

if >= then

continue

end if

// Evaluate the current optimization

// computation.

= ((V R) + Z(t))

if > then

= R

end if

end foreach