def selection_WST(data, t, tree=None):
# find all workers in MTD
MTD_RECT = np.array([[
t[0] - Params.ONE_KM * Params.MTD, t[1] - Params.ONE_KM * Params.MTD
], [t[0] + Params.ONE_KM * Params.MTD, t[1] + Params.ONE_KM * Params.MTD]])
locs = rect_query_points(data, MTD_RECT).transpose()
#locs = sorted(locs, key=lambda loc: distance(loc[0], loc[1], t[0], t[1]))
#u = 0
workers, dists = np.zeros(shape=(2, 0)), []
# find workers who would perform the task
for loc in locs:
dist = distance(loc[0], loc[1], t[0], t[1])
u_c = acc_rate(Params.MTD, dist)
#u = 1 - (1 - u) * (1 - u_c)
if is_performed(u_c):
workers = np.concatenate(
[workers, np.array([[loc[0]], [loc[1]]])], axis=1)
dists.append(dist)
# simulation
if workers.shape[1] == 0: # no workers
return 0, False, None, None, None, None, None
return len(locs), True, 0, random.choice(dists), 0, 0, 0
def geocast_knn(data, t):
# find all workers in MTD
# find all workers in the query
MTD_RECT = np.array([[t[0] - Params.ONE_KM * Params.MTD, t[1] - Params.ONE_KM * Params.MTD],
[t[0] + Params.ONE_KM * Params.MTD, t[1] + Params.ONE_KM * Params.MTD]])
locs = rect_query_points(data, MTD_RECT).transpose()
locs = sorted(locs, key=lambda loc: distance(loc[0], loc[1], t[0], t[1]))
u, dist, found = 0, 0, False
workers = np.zeros(shape=(2, 0))
for loc in locs:
workers = np.concatenate([workers, np.array([[loc[0]], [loc[1]]])], axis=1)
_dist = distance(loc[0], loc[1], t[0], t[1])
u_c = acc_rate(Params.MTD, _dist)
u = 1 - (1 - u) * (1 - u_c)
if is_performed(u_c):
if not found:
found = True
dist = _dist
if u >= Params.U:
break
# simulation
isPerformed, worker, dist_fcfs = performed_tasks(workers, Params.MTD, t, True)
hops_count, coverage, hops_count2 = hops_expansion(t, workers.transpose(), Params.NETWORK_DIAMETER)
if isPerformed: # the task is performed
return workers.shape[1], True, dist, dist_fcfs, hops_count, coverage, hops_count2
return workers.shape[1], False, None, None, hops_count, coverage, hops_count2
def selection_WST(data, t, tree=None):
# find all workers in MTD
MTD_RECT = np.array([[t[0] - Params.ONE_KM * Params.MTD, t[1] - Params.ONE_KM * Params.MTD],
[t[0] + Params.ONE_KM * Params.MTD, t[1] + Params.ONE_KM * Params.MTD]])
locs = rect_query_points(data, MTD_RECT).transpose()
#locs = sorted(locs, key=lambda loc: distance(loc[0], loc[1], t[0], t[1]))
#u = 0
workers, dists = np.zeros(shape=(2, 0)), []
# find workers who would perform the task
for loc in locs:
dist = distance(loc[0], loc[1], t[0], t[1])
u_c = acc_rate(Params.MTD, dist)
#u = 1 - (1 - u) * (1 - u_c)
if is_performed(u_c):
workers = np.concatenate([workers, np.array([[loc[0]], [loc[1]]])], axis=1)
dists.append(dist)
# simulation
if workers.shape[1] == 0: # no workers
return 0, False, None, None, None, None, None
return len(locs), True, 0, random.choice(dists), 0, 0, 0
# print np.sqrt(2*(1/0.1)**2), np.sqrt(((noisy - true) ** 2).mean())
for eps in eps_list:
true = np.genfromtxt("../log/true_count_KF_" + str(eps) + ".log", unpack=True)
noisy = np.genfromtxt("../log/noisy_count_KF_" + str(eps) + ".log", unpack=True)
true_flatten = true.flatten()
noisy_flatten = noisy.flatten()
p = Params(1000)
print "KF\t", "\t", np.sum(true_flatten), "\t", np.sum(noisy_flatten), "\t", np.sqrt(
2 * (T / (eps * (1 - p.structureEps))) ** 2), "\t", np.sqrt(((noisy_flatten - true_flatten) ** 2).mean())
true = np.genfromtxt("../log/true_count_KFPID_" + str(eps) + ".log", unpack=True)
noisy = np.genfromtxt("../log/noisy_count_KFPID_" + str(eps) + ".log", unpack=True)
true_flatten = true.flatten()
noisy_flatten = noisy.flatten()
p = Params(1000)
print "KFPID\t", "\t", np.sum(true_flatten), "\t", np.sum(noisy_flatten), "\t", np.sqrt(
2 * (T / (eps * (1 - p.structureEps))) ** 2), "\t", np.sqrt(((noisy_flatten - true_flatten) ** 2).mean())
x1 = time.time()
# for i in range(100):
max_dist = 100
dist = range(100)
Params.ZIPF_STEPS = 20
Params.s = 1
Params.MAR = 0.1
Params.AR_FUNCTION = "linear"
y = "\n".join(map(str, [acc_rate(max_dist, x) for x in dist]))
print y
print time.time() - x1
