scarf: a social-aware reliable forwarding technique for ... · a trusted vehicle is a vehicle that...
TRANSCRIPT
HAL Id: hal-01582503https://hal.archives-ouvertes.fr/hal-01582503
Submitted on 6 Sep 2017
HAL is a multi-disciplinary open accessarchive for the deposit and dissemination of sci-entific research documents, whether they are pub-lished or not. The documents may come fromteaching and research institutions in France orabroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, estdestinée au dépôt et à la diffusion de documentsscientifiques de niveau recherche, publiés ou non,émanant des établissements d’enseignement et derecherche français ou étrangers, des laboratoirespublics ou privés.
SCARF: A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications
Anna Vegni, Valeria Loscrí, Riccardo Petrolo
To cite this version:Anna Vegni, Valeria Loscrí, Riccardo Petrolo. SCARF: A SoCial-Aware Reliable Forwarding Tech-nique for Vehicular Communications. 3rd Workshop on Experiences with the Design and Implemen-tation of Smart Objects, MobiCom 2017, Oct 2017, Snowbird, United States. �hal-01582503�
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
SCARF: A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications
ANNA MARIA VEGNI, Department of Engineering
VALERIA LOSCRÍ, INRIA Lille-Nord Europe
RICCARDO PETROLO, Rice University
The Internet-of-Vehicles allows the coexistence of traditional Internet applications with the Internet-of-Things, so that vehicles cancommunicate not only among them but also to neighboring devices and humans as well. In this context, the human social behavior isa fundamental aspect that needs to be taken into account specially for message dissemination in Vehicular Social Networks.
In this paper, we present a probabilistic broadcast protocol that exploits the “social degree” of a vehicle to be selected as next-hopforwarder. The social degree of a vehicle depends on its social activity inside a vehicular social network, so that a vehicle that uses toaccess and post often in a given vehicular social network is more likely to be “socially trusted”. Selecting a trusted vehicle as next-hopforwarder allows a secure and faster message dissemination with respect to an “asocial” vehicle. Simulation results show that a sociallytrusted vehicle improves network performance, expressed in terms of overall throughput.
Additional Key Words and Phrases: Vehicular Social Networks, social degree, forwarding probability, collision probability
ACM Reference format:Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo. 2017. SCARF : A SoCial-Aware Reliable Forwarding Technique for VehicularCommunications. 1, 1, Article 1 (September 2017), 12 pages.https://doi.org/10.1145/nnnnnnn.nnnnnnn
1 INTRODUCTION
Nowadays, many vehicular applications are emerging, arising mainly from road safety and then to entertainment. Theconcept of Internet-of-Vehicles (IoV) [?] is then proliferating, and it is an inevitable convergence of the existing mobileInternet and the concept of Internet-of-Things (IoT). In IoV, Internet applications go on “wheels”, then converting theexisting vehicles into “digital cars” equipped with several technologies and sensors [?].
Moving from the existing concept of Vehicular Ad-hoc Networks (VANETs) [?] that allow opportunistic vehicularcommunications among vehicles i.e., vehicle-to-vehicle (V2V), and from vehicles to infrastructure i.e., vehicle-to-infrastructure (V2I), IoV technology refers to dynamic mobile communication systems that extend the concept ofvehicular communications to vehicle-to-human (V2H) and vehicle-to-sensor (V2S) interactions. It enables informationsharing and the gathering of information on vehicles, roads and their surrounds.
In IoV, the V2H interactions are governed by user behavior and needs. For instance, a user can book a ride to anunmanned taxi based on her needs and daily mobility. Then, understanding user behavior and mobility patterns isfundamental in the design of efficient communication systems in IoV. Also, real-life interactions among users have adirect impact on the performance of the network.
1
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
2 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
The dependence of mobility on social interactions has been a topic of interest of many researchers in the last years.Some of the first papers focusing on the strict relation between mobility model and social network theory has beeninvestigated in [?]. Musolesi and Mascolo [?] propose a new mobility model where the input of the model is the socialnetwork of the individuals carrying the mobile devices. By analyzing the results, the authors derive that the movementsof the handheld users are also driven by the social relationships of the individuals. This dependence has also beeninvestigated by Cho et al. in [?]. The authors state that the short-ranged travel is less impacted by the social networkstructure, while if a person travels a long distance then they are more likely to travel near an existing friend.
The meeting of mobility with social interactions rises to a particular class of social networks, namely the VehicularSocial Networks (VSNs) [?]. VSNs are online social networks where the social interactions are built on-the-fly, due tothe opportunistic links in vehicular networks. They exploit mobility aspects and basics of traditional social networks, inorder to create novel approaches of message exchange through the detection of dynamic social structures. For instance,a VSN can exist based on driver’s daily mobility i.e., moving every day from home to office. During the journey, thevehicle can access different social networks like (i) the network with members talking about traffic information i.e.,content-based social network, (ii) the network of a particular area of interest i.e., position-based social network, and (iii)the network with members belonging to the same office i.e., relationship-based social network. Then, we can figure outthat a vehicular social network is stronger impacted by short-ranged daily travels, than by long-ranged travels.
How social interactions affect message dissemination represents an important topic of interest for many researchersin the VSN context. Forwarding techniques in VSNs should not only consider existing constraints in vehicular ad hocnetworks (i.e., mobility, and connectivity issues), but also social aspects (i.e., messages are forwarded among trustedusers, which are sharing same interests and move in a common place at the same time). As known, traditional forwardingtechniques in VANETs are based on physical parameters, like the inter-vehicle distance so that the larger the distanceis, higher will be the forwarding probability [?]. Similar approaches based on this criterion have been presented, likethe Irresponsible Forwarding protocol which aims to limit the number of rebroadcasts [?]. However, when consideringVSNs where nodes share common interests with other neighboring vehicles, message forwarding techniques shouldconsider user social behavior and mobility pattern. For example, in [?] Herrera-Tapia et al. introduce a Friendly-Sharingdiffusion approach based on the consideration that the number of messages shared among nodes is improved when thecontact duration is increased.
In this paper, we present a probabilistic broadcast protocol based on the social activity of a vehicle in a vehicularsocial network. The proposed technique is named SoCial-Aware Reliable Forwarding (SCARF), which selects a next-hopvehicle according to its social parameter i.e., how much a vehicle is socially trusted inside a vehicular social network.Indeed, like in traditional online social networks like Facebook or Instagram, in a vehicular social network not all thevehicles are “social”, but it depends on the own social activity. For instance, a vehicle that uses to access a vehicularsocial network and posts comments every day on its own path from home to office is likely assumed to be a “social”vehicle, while a vehicle that sporadically connects to a vehicular social network and does not use to post is likely an“asocial” vehicle. Then, we can distinguish social vehicles inside a vehicular social network if their social (activity) degreeis higher than a given threshold.
Packet transmission via SCARF occurs on the basis of a probabilistic analysis i.e., the probability that a social vehiclewill forward a packet. The selection of a social vehicle as next-hop message forwarder is expected to affect networkperformances. Indeed, by comparing a social vehicle to an asocial one, we notice that due to its high social degree theformer can forward a message with a higher probability with respect to an asocial vehicle. As a result, by selecting
Manuscript submitted to ACM
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
SCARF : A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications 3
v1
v2d1
d2Tx pv1 = 0.43
pv2 = 0.6
Fig. 1. Schematic of the SCARF technique. At the same distance from a transmitter vehicle (i.e., d1 = d2), each potential forwardingvehicle can have a different social parameter, which corresponds to a different forwarding probability (i.e., pv1 , pv2 ).
social vehicles as potential forwarders we expect a more reliable message dissemination mechanism. This also allowssecure communications, since a social vehicle is assumed to be a trusted vehicle. Finally, in order to provide an effectivepacket transmission with no packet collisions, the SCARF protocol also supports a collision detection mechanism.
This paper is organized as follows. Section 2 describes the mathematical model of SCARF approach, by distinguishingthe forwarding probability and the collision probability. In Section 3 we report and discuss the simulation results relatedto SCARF technique. Finally, conclusions are drawn at the end of this paper.
2 PROPOSED TECHNIQUE
SCARF provides message rebroadcasting only to selected vehicles showing a social degree that makes them sociallytrusted. Indeed, SCARF’s main goal is to identify trusted vehicles that can forward messages in a fast and reliablemanner, thus allowing a high message dissemination rate within a VSN. This can be achieved while guaranteeing areduced collision probability. Thus, SCARF relies on two aspects i.e., the social-based reliable forwarding and the collisionprobability, respectively investigated in the following Subsection 2.1 and 2.2.
2.1 Social-based reliable forwarding
Our proposed technique is based on a social-based forwarding probability pf , so that a vehicle that receives a packetshould rebroadcast it only to the i-th “trusted” vehicle in its transmission range that has the following probability, i.e.:
pf ,i = exp[−ρ (z − di )
c · Si
], (1)
where di [m] is the inter-vehicle distance between the i-th receiving vehicle and the transmitting vehicle, ρ [veh/m] isthe exponentially distributed inter-vehicle spacing with mean value 1/ρ, c ⩽ 1 is a shape coefficient, and Si ⩽ 1 is thesocial parameter of the i-th vehicle.
A trusted vehicle is a vehicle that shows an appropriate social behavior with respect to a given VSN. For instance,a vehicle that accesses a VSN every day and has a high social activity degree (e.g., it posts several comments in theVSN) can be awarded as “socially trusted vehicle”. Mathematically, the social behavior of the i-th vehicle is expressed in
Manuscript submitted to ACM
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
4 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
terms of its social parameter Si ranging from 0 (i.e., the vehicle has no social behavior and then it is not socially trusted)to 1 (i.e., the vehicle has a social behavior, and then it is socially trusted). The social parameter takes into account thesocial degree of a vehicle in a given VSN, and it depends on (i) the degree of activity of a vehicle in a social network (i.e.,computed through the average number of posts and the number of accesses made), and (ii) the degree of correlation ofthe social behavior with respect to the mobility pattern.
Mathematically, we can define the social parameter as the probability that a vehicle is “socially active” in a VSN,assuming it is interested in the content discussed among members in the social network. For instance, let us assumetwo hypothesis (i.e., H1 and H2) respectively that (i) a vehicle is interested in the relevant content of a given VSN(namely, VSN1), and (ii) a vehicle is interested in another VSN (namely, VSN2). Then, under the hypothesis H1, thesocial parameter Si of the i-th vehicle is expressed as:
Si,H1 = Pr{ai,H1 > A���H1
}, (2)
where ai,H1 is the degree of social activity of the i-th vehicle depending on the average number of posts published inVSN1 (i.e., np,H1 ) and the average number of accesses in VSN1 (i.e., na,H1 ), i.e.
ai,H1 =np,H1
na,H1. (3)
In (2) the term A represents a threshold that determines if a vehicle has higher degree of social activity (i.e., if ai,H1 > A
or not).Figure 1 depicts the schematic of the SCARF technique. Let us consider a source vehicle (i.e., Tx vehicle) is transmitting
a message towards a next-hop vehicle inside its transmission range. Different vehicles can be eligible as next-hopforwarder, based on both physical metrics (i.e., the distance from the transmitter) and also on social parameters.Assuming two vehicles (i.e., v1 and v2) are at the same distance from the transmitter vehicle (i.e., d1 = d2 = 150 m), andthat the degree of (social) activity in VSN1 is a1,H1 = 0.5 and a2,H1 = 0.01 for v1 and v2, respectively, then a messagewill be forwarded to that vehicle having higher forwarding probability. Specifically, the probability of forwarding forv1 is pv1 = 0.43, while for v2 it is pv1 = 0.6. Vehicle v2 will be elected as socially-trusted vehicle, and then selected asnext-hop forwarder.
It follows that by using the Bayes theorem, (2) becomes:
Si,H1 =p1ai,H1
p1ai,H1 + p2ai,H2, (4)
where p1 and p2 are the probabilities that a vehicle is in the hypothesis H1 and H2, respectively. In (4), we considerai,H1 , ai,H2 . Figure 2 depicts the probability of having a social trusted vehicle to whom forward a packet. The vehicleis assumed to be connected in a given VSN (i.e., VSN1) that represents the scenario of hypothesis H1. The vehicle is thenconsidered as socially trusted if the probability (4) is higher than a given threshold (e.g., > 0.7). This occurs for differentvalues of aH1 and aH2 . For instance, according to Figure 2 for aH1 = 0.2 only the vehicle showing the lowest degree ofactivity in H2 (i.e., aH2 = 0.01) is considered as a social vehicle since its social parameter1 is higher than 0.7. Literallyspeaking, aH2 = 0.01 means that the degree of activity in VSN2 is very low and the vehicle is not socially-trusted inVSN2. As a consequence, for aH2 = 0.01 it is expected that the behavior of the vehicle in VSN1 will be higher and then,the probability that it is a socially-trusted vehicle in VSN1 will be higher.
1Notice that from (2) the social parameter represents a probability, and so the terms “probability of social vehicle” and “social parameter” of the i-thvehicle are given interchangeably in this paper.
Manuscript submitted to ACM
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
SCARF : A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications 5
0 0.2 0.4 0.6 0.8 1Degree of activity in H1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Pr
obab
ility
of s
ocia
l veh
icle
in H
1
a H2
= 0.01
a H2
= 0.25
a H2
= 0.5
a H2
= 0.75
a H2
= 1
Fig. 2. Probability of social-trusted vehicle versus the degree of activity under the hypothesis H1, for different values of the degree ofactivity in H2.
The social-based forwarding probability expressed in Eq. (1) is plotted in Figure 3 versus the degree of activity inhypothesis H1 (i.e., aH1 = [0, 1]) and for different values of inter-vehicle distance (i.e., d = [50, 150, 190] m) within thetransmission range of the source vehicle (i.e., d < 200 m), and of the degree of activity in hypothesis H2 (i.e., aH2 =
[0.01, 0.25, 0.5, 0.75, 1]). As previously said, the SCARF probability derives from traditional forwarding probabilistictechniques applied to vehicular communications, and then it depends on the inter-vehicle distance, so that the fartherthe forwarder from the source vehicle is, the higher the forwarding probability is. Thus, we observe that for increasingdistances the forwarding probability increases as well (see the red curves in Figure 3). However, the social feature istaken into account so that for a fixed distance not all the vehicles are potentially eligible as next-hop forwarder, but itdepends on the degree of social activity in H1 and H2.
Finally, in Figure 4 we show the social-based forwarding probability of a vehicle versus the inter-vehicle distanceinside the transmission range, for different values of the degree of activity in hypothesis H2. We notice that theforwarding probability increases for high values of the distance, and also for high values of the activity degree aH2 .This is due since the farthest the vehicle is in a transmission range, the highest the forwarding probability is; and also,the lowest the degree of activity in H2 is, the highest the forwarding probability is. For instance, if a potential next-hop
Manuscript submitted to ACM
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
6 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
0 0.2 0.4 0.6 0.8 1Degree of activity in H1
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Forw
ardi
ng p
roba
bilit
y in
H1
a H2
= 0.01 a H2
= 0.25 a H2
= 0.5 a H2
= 0.75 a H2
= 1
d = 190 m
d = 100 m
d = 50 m
Fig. 3. Probability of social-based forwarding versus the degree of activity under the hypothesisH1, for different values of inter-vehicledistance (i.e., d < z and z = 200 m) and degree of activity in H2.
vehicle lays at 150m from a source vehicle, based on the degree of activity in H2, the forwarding probability will changeaccordingly (e.g., pf = 0.31 for aH2 = 1, and pf = 0.6 for aH2 = 0.01).
2.2 Collision probability
The collision probability is the probability that at least two or more vehicles start transmitting at the same time. Thiscan be defined as
Pcoll = Pbusy (1 − Pt ), (5)
where Pbusy is the busy probability that at least one vehicle (i.e., the i-th vehicle) is using the channel at the same timeslot i.e.,
Pbusy = 1 −(1 − pf ,i
) (n−1), (6)
Manuscript submitted to ACM
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
SCARF : A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications 7
0 50 100 150 200Distance [m]
0
0.2
0.4
0.6
0.8
1Fo
rwar
ding
pro
babi
lity
in H
1 a H
2
= 0.01
a H2
= 0.25
a H2
= 0.5
a H2
= 0.75
a H2
= 1
Fig. 4. Probability of social-based forwarding versus the inter-vehicle distance (i.e., d < z and z = 200 m) under the hypothesis H1,for different values of degree of activity in H2.
and Pt is the probability that only one vehicle (i.e., the i-th vehicle) is using the channel i.e.,
Pt,i =pf ,i
(1 − pf ,i
) (n−1)Pbusy
, (7)
where n is the number of interfering vehicles.The trend of the collision probability is depicted in Figure 5 versus the inter-vehicle distance (i.e., d ≤ 200) for
different number of interfering vehicles (i.e., n = [2, 5, 10]) and the degree of activity in hypothesis H2. As expected,we observe that higher the number of interfering vehicles is, higher is the collision probability, as well as larger thenext-hop transmission range, higher the collision probability. Interesting we notice that for higher degree of activity inhypothesis H2, the collision probability is lower since in this case the forwarding probability is lower as well. However,we observe that for a lower degree of activity (e.g., see the solid red line for aH2 = 0.01 and n = 10) small increases ofthe distances correspond to very high values of collision probability. For instance, we get Pcoll = 0.143 for d = 2 m andPcoll = 0.354 for d = 4 m. As a remark, we observe that the collision probability changes the shape for small distances(i.e., < 50 m) and increasing interfering vehicles (i.e., solid red and blue curves), while for lower interferes this doesnot occur (see solid black line). This is expected since the scenario with high interfering vehicles in small distancesrepresents a likely collision event.
Manuscript submitted to ACM
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
8 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
0 50 100 150 200Distance [m]
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Colli
sion
prob
abili
ty
a H2
= 0.01 a H2
= 0.25 a H2
= 0.5 a H2
= 0.75 a H2
= 1
Fig. 5. Collision probability versus the inter-vehicle distance (i.e., d ≤ z and z = 200 m) under the hypothesis H1, for different valuesof interfering vehicles with this transmission range (i.e., n = [2, 5, 10] for black, blue, and red curves, respectively) and the degree ofactivity in H2.
By fixing a threshold for the collision probability, the next forwarding vehicle will be all those vehicles experiencinglower values for the collision probability. The collision threshold detects the maximum allowed distance from the sourcewhere a vehicle experiences the collision threshold. It varies according to VANETs factors, such as the inter-vehicledistance (i.e., d [m]) and the vehicular density (i.e., ρ [veh/m]), as well as the social factor i.e., the degree of activity inhypothesis H2. It is expressed as:
Thcoll = 1 − exp(−ρ · d
aH2
). (8)
Figure 6 depicts the dynamic trend of collision threshold versus the inter-vehicle distance for different degrees ofactivity. As expected, for a fixed distance, the collision threshold increases for decreasing degree of activity in H2.Indeed, as depicted in Figure 4, for lower values of aH2 the probability of forwarding will be likely higher than the caseof higher values of aH2 . It follows that the collision threshold is expected to be higher for lower degree of activity in H2,and vice versa. In Figure 6 we highlight two areas through a blue and a green rectangle, respectively indicating an areawith Thcoll ≤ 0.4 and Thcoll ≤ 0.5. In the first case (blue rectangle) the collision threshold is reached for a vehicle withManuscript submitted to ACM
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
SCARF : A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications 9
0 50 100 150 200Distance [m]
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1Co
llisio
n th
resh
old
a H2
= 0.01
a H2
= 0.25
a H2
= 0.5
a H2
= 0.75
a H2
= 1
Fig. 6. Collision threshold versus the inter-vehicle distance (i.e., d ≤ z and z = 200 m) under the hypothesis H1, for different valuesof degree of activity in H2.
aH2 = 1 at a distance of 25 m from the source vehicle. At the same distance for aH2 = 0.75 it corresponds a collisionthreshold of 0.5 (see the green rectangle). Again, the behavior of the collision probability can be observed in Figure 7varying the inter-vehicle distance specifically for (a) d = 50 m, (b) d = 150 m, and (c) d = 190 m for d < 200 m. Forincreasing distances the collision probability is higher, as well as for higher number of interfering vehicles the collisionprobability is higher. Finally, we observe that collisions can be affected also by the social parameter i.e., the degree ofactivity in H1 and H2. Indeed, as showed in Figure 3 a lower degree of activity in H1 corresponds to a lower forwardingprobability, and then a lower collision probability.
3 SIMULATION RESULTS
In this section, we show the simulation results expressed in terms of throughput experienced at the next-hop forwardervehicle. We assess the effectiveness of the SCARF technique with respect to a simple message dissemination protocolwithout social constraints. In this regard, we consider a scenario comprised of both socially-trusted (i.e., Si,H1 > 0.3with i = 1, . . . ,N ) and asocial (i.e., Sl,H1 ≤ 0.3, with l = 1, . . . ,N ) vehicles. All the vehicles are interested in the relevantcontent of a given VSN, namely VSN1. This corresponds to the hypothesis H1. Initially, we assume a packet arrival rateof 20, 50, and 100 packet/s, and a packet size of 1000 bytes.
The throughput has been computed for both the i-th socially-trusted and the l-th asocial next-hop vehicle laying at adistance of 150 m from the source vehicle (respectively, the black and blue lines in Figure 8). From Figure 8 we notice
Manuscript submitted to ACM
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
10 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
0 0.2 0.4 0.6 0.8 10
0.5
1
a H2
= 0.01 a H2
= 0.25 a H2
= 0.5 a H2
= 0.75 a H2
= 1
0 0.2 0.4 0.6 0.8 10
0.5
1
0 0.2 0.4 0.6 0.8 1Degree of activity in H1
0
0.5
1
���
���
���
Fig. 7. Collision probability under the hypothesis H1 versus the degree of activity in H1, for different values of interfering vehicleswith this transmission range (i.e., n = [2, 5, 10] for black, blue, and red curves, respectively) and the degree of activity in H2, in case of(a) d = 50 m, (b) d = 100 m, and (c) d = 190 m.
that, as expected, increasing the packet arrival rate corresponds to a higher throughput for the i-th socially-trustedvehicle, which reaches the maximum value for the highest degree of activity (i.e., aH1 = 1). On the other hand, for thel-th asocial vehicle the throughput trend is independent from the degree of activity. Indeed, since the vehicle is asocialtrusted, an increase of the degree of activity does not affect the probability of social-based forwarding as the SCARFtechnique selects only those vehicles with a social parameter higher than a given threshold (in our case, > 0.3).
Figure 9 depicts the throughput trend depending on (i) the packet arrival rate [pkt/s] (i.e., [1, 100]), (ii) the degree ofactivity in hypothesis H1 (i.e., 0 ≤ aH1 ≤ 1), and (iii) the inter-vehicle distance (i.e., d = [50, 100, 150] m), in case of (a)Manuscript submitted to ACM
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
SCARF : A SoCial-Aware Reliable Forwarding Techniquefor Vehicular Communications 11
0 0.2 0.4 0.6 0.8 1Degree of activity
0
0.05
0.1
0.15
0.2
0.25
0.3Th
roug
hput
[MBi
t/s]
pkt arrival rate = 20pkt arrival rate = 50pkt arrival rate = 100
Fig. 8. Comparison of throughput [MBit/s] for a socially-trusted vehicle (black lines) and an asocial vehicle (blue lines), versus thedegree of activity in H1 and different values of packet arrival rate, for a fixed distance of d = 150 m.
socially-trusted, and (b) asocial next-hop vehicle. We notice that the highest values of throughput are experienced for asocially trusted vehicle (see Figure 8 (a)). In both cases, we observe that highest trend is experienced for the longestinter-vehicle distance (i.e., d = 150) and the maximum value is obtained for the highest degree of activity and packetarrival rate. This corresponds to the higher forwarding probability for a vehicle that lays longer from the source andhas higher degree of activity.
4 CONCLUSIONS
In this paper we presented SCARF, a technique for message forwarding in VSNs. SCARF is based on the concept thatnot all the vehicles have the same social degree in a VSN, so that a few vehicles are considered as “socially-trusted”,while others as “asocial”. SCARF selects the most socially-trusted vehicle as next-hop forwarder, in order to guaranteereliable communications, while also guaranteeing a collision-aware mechanism.
Manuscript submitted to ACM
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
12 Anna Maria Vegni, Valeria Loscrí, and Riccardo Petrolo
Packet arrival rate [pkt/s] Degree of activity
0
0.1
0.2
0
0.3
0.4
0.5
Thro
ughp
ut [M
Bit/s
]
0.6
0.7
25 150 0.750.575 0.25100 0
0.1
0.2
0.3
0.4
0.5
0.6
d = 50
d = 100
d = 150
(a)
Degree of activity Packet arrival rate [pkt/s]
Thro
uhgh
put [
MBi
t/s]
0
0.1
0
0.2
0.3
0.4
0.5
25 10.7550 0.575 0.25100 0
0.5
1
1.5
2
2.5
3
3.5
4
4.5 d = 150
d = 100
d = 50
(b)
Fig. 9. Throughput [MBit/s] versus the degree of activity in H1 and the packet arrival rate [pkt/s], for different values of inter-vehicledistance, in case of (a) socially-trusted, and (b) asocial next-hop vehicle.
Manuscript submitted to ACM