#StandardsGoals for 2024: What’s new for BISAC - Tech Forum 2024
Phd Report 2017
1. Dottorato di Ricerca in Ingegneria dell’Informazione
Mattia G. Campana
2nd YEAR ACTIVITY REPORT, CICLO XXXI
Tutors: Ing. Enrico Gregori, Ing. Franca Delmastro
2. RECOMMENDER SYSTEM FOR
OPPORTUNISTIC ENVIRONMENTS
RESEARCH TOPIC
Mattia G. Campana 1PhD Activity Report 2017
Recommender System (RS): estimate a utility function that automatically predicts
how much a user will like an item
RS for OE
Device-to-device wireless communication (D2D)
Opportunistic Environment (OE)
Human mobility
Store-carry-forward paradigm
Optimize routing/dissemination of info/data
Discover data from devices in proximity for
high-level applications (e.g., Mobile Social Networks - MSN)
3. Mattia G. Campana 2PhD Activity Report 2017
Tags
Items
Users
PLIERS - RECOMMENDER SYSTEM
1st YEAR ACTIVITY
Proposed for OSN in which users associate tags to their
content in order to describethem from a semantic point of view
PLIERS models the users-items-tags relationships
as a tripartite graph
folksonomy ≠ ontology
no relationships between tags
Using a diffusion process, PLIERS generates a ranking
of items previously unseen by the target user’s
Recommends the top-n items
4. Mattia G. Campana 4
PERVASIVE-PLIERS
A framework proposed for highly distributed environments (e.g., MSN)
Each device builds its own local knowledge graph (LKG)
Update LKGs
Fetch relevant items
Info about items created by its local user
+
local knowledge of other encountered nodes
Limits exchanged data and optimize the content dissemination in D2D network
1st YEAR ACTIVITY
PhD Activity Report 2017
5. Mattia G. Campana 5
2nd YEAR ACTIVITY
SURVEY ON RS 4 OSN & MSN
Social RS
Social-aware
Tag-based
Location-based
Friendships relations
Followers / Followee relations
Trust relations
(User-defined) Tags
Location (POIs and trajectories)
Time
Locations’ meta-information (e.g., tags)
Social & Trust relations
People
Items
Tags
Locations
RS FOR OSN
RS FOR MSN
RESEARCH DIRECTIONS
We compared the few solutions proposed in the literature based on their rec. model (CF and tag-based), the set of
context information used (ratings, location, tags) and their evaluation process.
Introduce additional info specifically related to the mobile environment (e.g., mobility, sensors data, apps usage)
Characterize the user’s social context using info coming from both virtual and physical world (e.g., OSN + physical contacts)
Need of a common framework and appropriate datasets to evaluate RS for MSN
PhD Activity Report 2017
6. Mattia G. Campana 6
2nd YEAR ACTIVITY
WI-FI DIRECT - GROUP MANAGER
A novel communication protocol to enable D2D communication
among commercial mobile devices
Exploits context information to autonomously setup the communication
It is based on Wi-Fi Direct standard
PhD Activity Report 2017
7. GO
WI-FI DIRECT (WFD)
GENERAL OVERVIEW
In Wi-Fi Direct (WFD) nodes can communicate to each other only
if they belong to the same WFD Group (star topology)
Group Owner (GO) is the “leader” of the group.
It implements the functionalities of a IEEE 802.11 Access Point (AP)
Client
Clients: both WFD-enabled and “legacy” devices
see a GO as a traditional AP
Two WFD Groups in proximity can not communicate to each other
Mattia G. Campana 7PhD Activity Report 2017
8. PEER
DISCOVERY
SERVICE
DISCOVERY
WFD
HOW TO DISCOVER DEVICES IN PROXIMITY
IEEE
802.11
scan
Probe req.
Search
ch. 1,6,11
Search
ch. 1,6,11
Listen
ch. 6
Listen
ch. 11
Search
ch. 1,6,11
Search
ch. 1,6,11
Listen
ch. 6
Probe
resp.
D2 found
Probe req. Probe req.
Probe req.
D1
D2
PEER
DISCOVERY
D2 found
GAS Initial
Request
GAS Initial
Response
Received D2
Services
D1
D2
Mattia G. Campana 8PhD Activity Report 2017
9. WFD
GROUP FORMATION
STANDARD
AUTONOMOUS
Mattia G. Campana 9
PEER
DISCOVERY
WPS DHCPGO
negotiation
response
GO
negotiation
request
GO
negotiation
confirm
D1
D2
D1
WPS DHCP
IEEE
802.11
scan
D2
Nodes send a GO Intent (GI) value, which
represents their willingness to become GO
PhD Activity Report 2017
10. Mattia G. Campana 10
WFD / IMPLEMENTATION
LIMITATIONS
Accept connection
from
Device_XYZ ?
GO Intent is not related to the suitability of a node to act as GO.
(It is a random value or set by applications).
Peer discovery + GO Negotiation may require several seconds
WPS requires manual user’s authorization (PIN or Accept button).
PhD Activity Report 2017
11. Mattia G. Campana 11
WFD-GM
PROPOSED SOLUTION
We propose Wi-Fi Direct - Group Manager (WFD-GM), a novel protocol for the efficient configuration and
management of WFD groups to enable opportunistic networks with real commercial devices.
Combines two mechanism of WFD standard to identify the
best group configuration:
Avoids the manual user’s authorization
Service Discovery
Autonomous Group Formation
Enables the content/information diffusion among different WFD groups
Does not require any modification of O.S. or WFD standard
PhD Activity Report 2017
12. Mattia G. Campana 12
WFD-GM
INITIALIZATION
Each node creates a WFD
group electing itself as GO
(Autonomous Group Formation)
Shares the group credentials
among nodes in proximity
(Service Discovery)
GOAL: Speed up the group formation and the credential exchange
PhD Activity Report 2017
13. Mattia G. Campana 13
WFD-GM
CONTEXT INFORMATION
Indicates the suitability of the local node to become GO of a larger group.
It is a linear combination of the following set of context features:
available resources (e.g., battery level, free CPU/memory)
# current peers in proximity (LN)
# incoming connections that the device can still accept
how much faster LN changes over time (Stability Index)
Bad GO
Bad GO: LN changes quickly Good GO: LN changes slowly
(Its group will be rapidly destroyed) (It is able to create a long-lasting group)
In addition to the group credentials, each node shares its Suitability index (si)
PhD Activity Report 2017
14. Mattia G. Campana 14
WFD-GM
NODE STATUS
GO1: node has no clients but LN is not empty (nearby nodes)
GOElection Procedure:
My si = max si ?
remain GO and wait for incoming connections
connect as legacy client to the GO with the max s(ln)
Yes
No
GO2: node has some clients but the
amount of resources consumed to
manage the current group is beyond
a predefined threshold resth
n5
n1
n3
n4
n2
n5
n1
n3
n4
n2
It destroys the group and comes back to the initial status GO1
n3
n5
n1
n4
n2
Every TD seconds (decision time), each node check its status which can be one of the following:
PhD Activity Report 2017
15. Mattia G. Campana 15
GO has discovered
another GO in proximity
Based on their suitability
indices, it is the best GO
GO asks to its clients if
they “see” the other GO
If the majority agree, GO
disbands its group and
connects to the new one
Best GO
WFD-GM
NODE STATUS
GO3 (merge procedure): node evaluates to merge its group with another one in proximity.
C1 (traveler procedure): a client has discovered another GO in proximity.
With probability pT it
becomes a traveler
Node blacklists the old GO
for a fixed amount of time
Node choose which group to
connect among those in proximity
PhD Activity Report 2017
16. Mattia G. Campana 16
WFD-GM
MIDDLEWARE-LAYER PROTOCOL
Configuration & Management
of WFD Groups
Security policies to control
the communication among
nodes (e.g., trust
management and
cooperation protocols,
apps/user specific privacy
protection)
Operating System
Context
Manager
NET MANAGER
APPS MANAGER
APP
1
APP
2
APP
N-1
APP
N
..…
WFD-GM
ROUTING / DATA
DISSEMINATION
SECURITY-PRIVACY
Battery level, CPU/memory usage,
nodes in proximity, …
PhD Activity Report 2017
17. Mattia G. Campana 17
EVALUATION
EXPERIMENTS SETUP
0 10 20 30
Hour
0
0.2
0.4
0.6
0.8
1
Batterylevel
Group size: 2
Group size: 20
Intermediate
0 10 20 30
Hour
0
0.2
0.4
0.6
0.8
1
Batterylevel
Group size: 2
Group size: 20
Intermediate
We compared WFD-GM with a Baseline protocol
GO election: node with the highest MAC addr.
The GO maintains its role until the end of its resources or in case of out-of-range
We implemented both WFD-GM and Baseline in the ONE opportunistic simulator
Predicted battery depletion
Parameters estimation with real commercial devices
Battery depletion
GO w/o clients + Service Discovery every 2min: 20% every 5h
Limited number of clients: rand(4,15) for each node e.g., LG Nexus 5 and Motorola Nexus 6: 4 clients
HTX Neus 5X and Xiaomi Mi5: 10+ clients
GOs Clients
Each member of the group ([1,4] clients) sends a message to
the others every 100ms (worst case scenario). Then, we used a
linear regression model to estimate the power consumption in
larger groups.
PhD Activity Report 2017
18. Mattia G. Campana
EVALUATION
SIMULATED SCENARIOS
18
ComiCon
# nodes: 2000
Mobility: [0,1.5] m/s - ShortestPath
575 POIs (e.g., stands, eateries)
Each node waits from 10min to 1h at
each POI (e.g., queues)
Time: 4 h
Helsinki
# nodes: 4000
Mobility: Working Day Mobility
Model
Time: 24 h
Concert
# nodes: 1000
Mobility: fixed positions
Time: 3 h
Main Stage
We simulated 3 application scenarios with different numbers of nodes (users) and different mobility patterns
PhD Activity Report 2017
19. Mattia G. Campana
EVALUATION
MESSAGE DIFFUSION
0
0.2
0.4
0.6
0.8
1
0 0.5 1 1.5 2 2.5 3
innodes’caches(%)
Hour
Baseline 5
Baseline 30
Baseline 60
WFD-GM 5
WFD-GM 30
WFD-GM 60
0
0.2
0.4
0.6
0.8
1
0 0.5 1 1.5 2 2.5 3 3.5 4
Meannumberofmessages
innodes’caches(%)
Hour
Baseline 5
Baseline 30
Baseline 60
WFD-GM 5
WFD-GM 30
WFD-GM 60
0
0.2
0.4
0.6
0.8
1
0 5 10 15 20 25
Meannumberofmessages
innodes’caches(%)
Hour
Baseline 5
Baseline 30
Baseline 60
WFD-GM 5
WFD-GM 30
WFD-GM 60
Concert Comicon Helsinki
19
When a simulation starts, each node generates a message
We assume that nodes implement an epidemic forwarding algorithm
When a node joins a WFD group, it sends all the messages contained in its own cache to all the members of the group
Every 30 minutes (sim. time), we measured the % of message contained in the nodes’ caches
PhD Activity Report 2017
20. Mattia G. Campana 20
EVALUATION
CONNECTIVITY GRAPH
Both Baseline and WDF-GM create a network of multi-hop paths among the nodes, called Connectivity Graph (CG)
In CG, two nodes are connected if they have participated in the same WFD group
n5
n1
n3
n4
n2
n1
n2n3
n4n5
WFD Group Corresponding CG
Total connection time
PhD Activity Report 2017
22. Mattia G. Campana
EVALUATION
NETWORK CONNECTIVITY & RESOURCES
22
0
20
40
60
80
100
Concert Comicon Helsinki
Finalbatterylevel
6%
9%
Times at which nodes expire their
batteries (i.e., 71% of the sim. time)
WFD-GMBaseline
0
3
6
8
11
5 30 60 5 30 60 5 30 60
80
87
93
100
Concert Comicon Helsinki
99
9999
99
99
99
100 100 100100 100
#ofCG’sconnectedcomponents
2
% nodes in the largest
connected component
100
2 2
PhD Activity Report 2017
23. Mattia G. Campana
COURSES
FORMATION ACTIVITY
23
Academic Writing and Academic Presentation skills (4 CFU)
PhD+ 2016 (9 CFU)
Signal processing and mining of Big Data: biological data as case study (5 CFU)
Deep Learning (3 CFU) - External
1st YEAR
Massive MIMO - Fundamentals and state-of-the-art (4 CFU)
Fuzzy Logic and Fuzzy Systems (3 CFU)
ACM Summer School on Recommender Systems 2017 (9 CFU) - External
2nd YEAR
TOTAL CFU: 37
PhD Activity Report 2017
24. Mattia G. Campana
PUBLICATIONS
24
[J1] Valerio Arnaboldi, Mattia Giovanni Campana, Franca Delmastro, and Elena Pagani. A Personalized Recommender System for
Pervasive Social Networks. Pervasive and Mobile Computing 36 (2017): 3-24.
[C1] Valerio Arnaboldi, Mattia Giovanni Campana, Franca Delmastro, and Elena Pagani. 2016. PLIERS: a popularity-based
recommender system for content dissemination in online social networks. In Proceedings of 31st Annual ACM Symposium on
Applied Computing (ACM SAC 2016).
[C2] Mattia Giovanni Campana, Franca Delmastro, and Raffaele Bruno. 2016. A Machine-Learned Ranking Algorithm for Dynamic
and Personalised Car Pooling Services. In Proceedings of the 19th International Conference on Intelligent Transportation Systems
(IEEE ITSC 2016).
1st YEAR
2nd YEAR
[J2] Mattia Giovanni Campana, Franca Delmastro. Recommender Systems for Online and Mobile Social Networks: a survey. Online
Social Networks and Media. [ UNDER REVIEW - Minor revision ]
[C3] Valerio Arnaboldi, Mattia Giovanni Campana, Franca Delmastro: 2017. Context-Aware Configuration and Management of WiFi
Direct Groups for Real Opportunistic Networks. In Proceedings of the 14th IEEE International Conference on Mobile Ad-hoc and
Sensor Systems (IEEE MASS 2017).
PhD Activity Report 2017