1. Introduction
Opportunistic Content Caching
Vehicular Mobility Role in Cooperative Content Caching
Conclusions and Future Work
Opportunistic and Cooperative Content Caching
Paradigms in Wireless Networks
Osama Gamal Mohamed Attia
Wireless Intelligent Networks Center
School of Communication and Information Technology
Nile University, Egypt
August 5, 2012
1 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
2. Introduction
Opportunistic Content Caching
Vehicular Mobility Role in Cooperative Content Caching
Conclusions and Future Work
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
2 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
3. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
3 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
4. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Content Caching
Content caching has been introduced in the Internet, and later for wireless
extensions, to enhance user experience (retrieval time) and reduce network
load.
It allows nodes to store a copy of the data it do request in a previous time slot
for a future use.
Different caching paradigms emerged in MANETs:
Non-cooperative: nodes make independent decisions to cache data or paths.
Cooperative: exploits the wisdom of the crowd and creates diversity.
Opportunistic: utilizes the data sent in the network for future requests.
4 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
5. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
5 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
6. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Main Contribution
In the first part:
1 We introduce the novel concept of OCC whereby nodes cache overheard
content delivered by the content server (CS) to nearby nodes.
2 We cast the OCC problem into a mathematical framework inspired by the
diversity-multiplexing tradeoff first introduced by David Tse.
3 We characterize the diversity gain of OCC and quantify the improvement over
a baseline which does not leverage the inherent broadcast nature of wireless
transmissions
6 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
7. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Main Contribution
Continue ..
In the second part:
1 Introduce a definition for the Probability of Outage in the context of
cooperative content caching.
2 Characterize, analytically, the outage probability under vehicular and random
mobility scenarios.
3 Compare, using simulations, the outage performance under sample mobility
regimes.
7 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
8. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
8 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
9. Introduction
Background
Opportunistic Content Caching
Main Contribution
Vehicular Mobility Role in Cooperative Content Caching
Related Work
Conclusions and Future Work
Related Work
Content caching concept has been first introduced to the Internet, especially
for web [Wang ’99, Barish et al. ’00].
Cooperative Content Caching in MANETs:
Yin et al., 2006: proposed three schemes for cooperative caching in ad hoc
networks with the objective of reducing the query delay.
Fiore et al., 2009: introduced a new metric (presence index) deciding for how
long should a data chunk be cached.
El Gamal et al. 2010: introduced novel proactive resource allocation scheme
and analyzed it under DMT framework.
Fiore et al., 2007: studied the impact of highway and urban mobility on
VANET routing protocols.
9 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
10. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
10 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
11. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Assumptions
Time slotted system of a single content server (CS)
and multiple nodes.
For any node i, let N average number of nodes
randomly dispersed within the CS radio range. i
A file is composed of m fixed number of chunks.
Chunk requests arrive at an arbitrary node i in each
slot according to a Poisson process with rate λi = λ.
Requests arrive at the beginning of a slot and each
chunk is retrieved in one slot using one resource
(channel).
Node i has wireless capacity with total number of
channels C. Delivering requested content to
node i
Node’s cache size is M chunks, M >> C (supported Cache overheard content
delivered to node i for T slots
by Moore’s law).
11 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
12. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Opportunistic Content Caching Scheme
Start
All nodes run in promiscuous
passive mode.
Overhear transmitted data
chunks between the
Node i overhears and stores or Content Server (CS) and
nearby nodes
updates the new content sent from
the Content Server to any of the
nearby nodes.
No Encounter a new
Node i caches new, or updated, data chunk?
overheard data chunks for T time Yes
slots.
Does it exist in No
Given the overlap in interests, p, a current cache? Cache for T time slots
query issued by node i may be
Yes
served from its own cache or from
Update if newer
the content server (0 ≤ p ≤ 1).
12 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
13. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
13 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
14. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Diversity-Multiplexing Tradeoff Mathematical Framework
Originally proposed by David Tse et al. for multi-antenna wireless
communication.
DMT allows analyzing the asymptotic decay rate of outage probability with
the system capacity C.
We assume that the total request arrival rate per slot λ scales with capacity in
two different regimes:
Linear Scaling: λ = γC
Polynomial Scaling: λ = Cγ
where γ serves as the bandwidth utilization factor, 0 ≤ γ ≤ 1
As γ goes to 1, the system becomes critically stable and more subject to outage
events.
14 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
15. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Definition of Outage
Definition
We define the probability of outage at any arbitrary node as the probability of not
being able to serve a request within a time slot.
In this case, Opportunistic Caching, an outage event takes place when a node is
not being able to retrieve a requested data chunk, in a given time slot, from the
content server, or the cached data overheard from chunk retrievals of nearby
nodes.
15 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
16. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Diversity Gain
Overhearing and caching data chunks retrieved by nearby nodes from the
Content Server yields multi-user diversity.
Overlapping requests may be resolved locally using the overheard data cached
from prior deliveries to the N nearby nodes, at no cost versus download from
the content server at a delay and delivery cost.
We define diversity gain under as follows:
1 Linear Scaling:
log P(O)
d(γ) = lim −
C→∞ C
2 Polynomial Scaling:
log P(O)
d(γ) = lim −
C→∞ C log C
16 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
17. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
17 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
18. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Retrieval from Content Server Only (Baseline)
All the requests will be served by the content server with no provisions for
caching or cooperation among the nodes.
The probability of outage, P(O) , will be only the outage at server:
P(O) = Pcs (O)
Let Q(n) be the number of requests at a node in the time slot n. We can express
the probability of outage as follows:
P(O) = P(Q(n) > C)
∞
e−λ λk
=
k!
k=C+1
18 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
19. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Diversity Gain of Baseline Retrieval
From the previous equations, we can rewrite the diversity gain in case of linear
capacity scaling as follows:
1
dbl (γ) = − lim log P(Q(n) > C)
c→∞ C
Based on the analysis by El Gamal et al., it can be shown that the diversity
gain of the baseline no caching system, in case of linear capacity scaling, is
given by,
dbl (γ) = γ − 1 − log γ
Also, in case of polynomial capacity scaling we can write the diversity gain as:
1
dbl (γ) = − lim log P(Q(n) > C)
c→∞ C log C
=1−γ
19 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
20. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
20 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
21. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Opportunistic Content Caching (OCC)
In this paradigm, each node has a cache storage that hosts data overheard from
nearby nodes within the past T time slots.
The outage probability is the probability of not finding the requested chunk in
the cached overheard data and not being able to retrieve it from the Content
Server due to the limited wireless capacity, C, that is,
P(O) = Pcs (O)Poh (O)N
Poh (O) is the probability of not being able to resolve the query from the
cached overheard data.
The outage probability Poh (O) equals to the probability that a node of the N
nearby nodes didn’t make any overlapping requests within the last T time slots.
21 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
22. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Opportunistic Content Caching (OCC)
Continue ..
Poh (O) can be written as follows
Poh (O) =[P(O|Q(n) ≤ C)P(Q(n) ≤ C) + P(O|Q(n) > C)P(Q(n) > C)]T
We know that, the outage probability given the number of requests is less than
or equal C equals to e−pλ which is the probability of not finding overlapping
requests.
Also, the probability of outage given the number of requests greater than C is
guaranteed to be equal to 1. Hence,
T
Poh (O) = e−pλ P(Q(n) ≤ C) + P(Q(n) > C)
T
= e−pλ + (1 − e−pλ )P(Q(n) > C)
22 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
23. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Diversity Gain of Opportunistic Content Caching
Continue ..
By substituting in the outage definition and taking the logarithm:
log P(O) =TN log e−pλ + (1 − e−pλ )P(Q(n) > C) + log P(Q(n) > C)
Simplifying and solving for the linear scaling, we find that,
dopp (γ) = TN min(pγ, dbl (γ)) + dbl (γ)
So, if there is no overlapping requests between nodes (i.e. p = 0), we find out
that dopp (γ) = dbl (γ).
However, at the total overlap between nodes’ requests (i.e. p = 1), it is clear
that dopp (γ) = (TN + 1)dbl (γ).
Also, solving for the polynomial scaling case show that no improvement over
baseline-retrieval:
dopp (γ) = dbl (γ) = 1 − γ
23 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
24. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
24 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
25. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Results and Insights
We show the result of the probability of outage under the linear and
polynomial capacity scaling cases.
WLOG, We plotted the curves at an arbitrary values listed in the table below in
order to show the improvement of the opportunistic content caching over the
baseline scenario.
Parameter Value
Multiplexing gain (γ) 0.75
Interest overlap probability (p) 0.6
Number of neighbor (N) 3 nodes
Caching time (T) 6 slots
25 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
26. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Outage probability performance
0 0
10 10
−10
10
−10
10
Outage Probability P(O)
Outage Probability P(O)
−20
10
−20
10
−30
10
−30
10
−40
10
Baseline Baseline
Opportunistic Opportunistic
−40 −50
10 10
0 10 20 30 40 50 60 0 10 20 30 40 50 60
Capacity (C) Capacity (C log C)
(a) Linear Scaling Case (b) Polynomial Scaling Case
26 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
27. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
Diversity-Multiplexing Tradeoff
5 1
Baseline Baseline
4.5 Opportunistic 0.9 Opportunistic
4 0.8
3.5 0.7
3 0.6
Diversity
Diversity
2.5 0.5
2 0.4
1.5 0.3
1 0.2
0.5 0.1
0 0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Multiplexing Gain (γ) Multiplexing Gain (γ)
(c) Linear Scaling Case (d) Polynomial Scaling Case
No improvement in terms of diversity gain for the polynomial scaling case. This
could be justified since the content caching scheme under polynomial scaling with
γ
an overlapping factor that grows as e−pC is very slow.
27 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
28. System Model
Introduction
DMT Framework for Opportunistic Content Caching
Opportunistic Content Caching
Baseline Retrieval
Vehicular Mobility Role in Cooperative Content Caching
Opportunistic Caching
Conclusions and Future Work
Results and Insights
The effect of interest overlap probability
0 0
10 10
−10
−5 10
10
Outage Probability P(O)
Outage Probability P(O)
−20
10
−10
10
−30
10
−15
10
−40
10 Baseline
Baseline
Opportunistic (p = 0.1) Opportunistic (p = 0.1)
−20
10 Opportunistic (p = 0.4) −50 Opportunistic (p = 0.4)
10
Opportunistic (p = 0.7) Opportunistic (p = 0.7)
Opportunistic (p = 1) Opportunistic (p = 1)
−25 −60
10 10
0 2 4 6 8 10 12 14 16 18 20 0 10 20 30 40 50 60
Capacity (C) Capacity (C log C)
(e) Linear Scaling Case (f) Polynomial Scaling Case
28 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
29. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
29 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
30. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Motivation
Vehicular Ad hoc Networks (VANET) is a promising emerging networking
paradigm.
VANETs are envisioned to improve the driving experience and save lives on
the roads.
Cooperative content caching (CCC) is a plausible technology for content
delivery in VANETs.
Content delivery to mobile platforms, e.g., vehicles, from infrastructure is
resource- and time-consuming. Hence, cooperation presents an opportunity.
Is there a performance gain for the vehicular mobility over random mobility.
If there is a performance gain, how to quantify it?
30 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
31. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
31 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
32. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
System Model
We assume toy model of two nodes (adequate to
capture the problem).
Users are interested in items where each information
Y
item consists of multiple chunks.
Direction of
Nodes starts with empty caches. Movement
Chunk requests arrive at node i according to a l
r
x
Poisson process with rate λ.
xx
θ X
Fixed transmission power which translates to a n1 n2
circular range of radius r.
If the requesting node gets a query resolved, it caches
a copy of the chunk for an arbitrarily long time.
32 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
33. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Mobility Models
Random Mobility:
x: Distance between the two Y
vehicles, x ∼ Uni[−r, r].
v: Relative velocity, Direction of
Movement
v ∼ Uni[vmin , vmax ].
θ: Direction of movement,
θ ∼ Uni[θmin , θmax ].
Vehicular Mobility: l
r
x
x ∼ Uni[−r, r].
v ∼ Uni[vmin , vmax ]. xx
θ X
Direction of movement is
deterministic, θ = π/2 for a n1 n2
straight freeway segment.
33 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
34. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
34 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
35. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Probability of Outage
Definition
We define the probability of outage, Pn1 , as the probability of not finding a data
o
chunk at a single-hop neighbor within time period (t, t + τ ).
Pn1 can be defined as the complement of the probability of node n1 finding a
o
chunk, denoted Pn1 .
f
The event of finding a data chunk happens when 3 independent events jointly
take place:
n2 requests at least a chunk within the period τ .
There is an interest overlap with probability γ.
The two nodes are within the communication range (Pneigh ).
Pn1 = 1 − Pn1
o f
= 1 − γ(1 − e−λτ )Pneigh
35 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
36. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Quantifying Pneigh in Random Mobility
n2 will stay within the radio range of n1 after
time τ iff if vτ is less than or equal to
distance l.
l= 1 − x2 sin2 θ − x cos θ
Hence,
Pneigh = P(vτ ≤ l)
= P(vτ ≤ 1 − x2 sin2 θ − x cos θ)
= f (x, u, θ)dx du dθ
x,u,θ∈Dr
The integration is solved numerically due to
its complexity.
36 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
37. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Quantifying Pneigh in Vehicular Mobility
In this case θ = π/2, and,
√
l = 1 − x2 . Hence,
Pneigh = P(τ vmin ≤ u ≤ min(τ vmax , l))
= f (x, u)dxdu
x,u∈Dv
umax
√
1 − u2
= du
umin umax − umin
Dv is the region over which x and u
satisfy the inequality:
τ vmin ≤ u ≤ min(τ vmax , 1 − x2 )
37 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
38. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Outline
1 Introduction
Background
Main Contribution
Related Work
2 Opportunistic Content Caching
System Model
DMT Framework for Opportunistic Content Caching
Baseline Retrieval
Opportunistic Caching
Results and Insights
3 Vehicular Mobility Role in Cooperative Content Caching
Motivation
System Model
Outage Performance Analysis
Performance Results
4 Conclusions and Future Work
38 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
39. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Simulation Settings
We develop Matlab simulations to verify the analytical results.
Analytical and simulation results are generated using the following system
parameters:
Parameter Value
Overlap ratio (γ) 0.7
Requests arrival rate (λ) 3 requests/sec
Radio range (r) 150 m
Minimum relative speed (vmin ) 5 km/hr
Maximum relative speed (vmax ) 50 km/hr
39 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
40. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Performance Results
1 1
Random Mobilty (Analysis)
0.9 Vehicular Mobility (Analysis) 0.9
)
Random Mobility (Simulation)
neigh
0.8
Vehicular Mobility (Simulation)
Probability of being in reach (P
0.8
0.7
Probability of Outage
0.6 0.7
0.5
0.6
0.4
0.3 0.5 Random Mobilty (Analysis)
Vehicular Mobility (Analysis)
0.2
0.4 Random Mobility (Simulation)
0.1 Vehicular Mobility (Simulation)
0.3
0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
τ (sec) τ (sec)
(g) Probability of being in reach (h) Outage Probability
For the range of Po of practical interest, vehicular mobility has lower probability of
outage than random mobility.
40 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
41. Introduction Motivation
Opportunistic Content Caching System Model
Vehicular Mobility Role in Cooperative Content Caching Outage Performance Analysis
Conclusions and Future Work Performance Results
Performance Results
Continue..
Comparing random mobility to road width-limited vehicular mobility (5-lane
freeway with 4 meters lane width).
1
0.9
0.8
Probability of Outage
0.7
0.6
0.5
0.4 Random Mobility
Vehicular Mobility
0.3
0 20 40 60 80 100 120
τ (sec)
Confirms the superiority of vehicular mobility especially in the practical range of
interest.
41 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
42. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Conclusions
In the first part: Then, w Follows,
We proposed a new paradigm for content caching that involves exploiting the
prior resolved queries of the neighbor users for future requests.
We formally set forth the definition of outage event in lights of a plausible
system model.
We conducted diversity-multiplexing tradeoff analysis (diversity as chances of
resolving queries in terms of number of nodes and time slots).
We evaluated, mathematically, the outage probability and diversity gains of the
system under different settings.
Finally, numerical results that validate our claims are shown and insights are
drawn.
42 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
43. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Conclusions
Continue ..
In the second part:
We introduced a formal definition for the probability of outage in the context
of cooperative content caching.
Then, we characterized, analytically, the outage probability under vehicular
and random mobility.
We verified the analytical results using simulation studies which exhibit
complete agreement.
Results confirm the opportunity created by the structured vehicular mobility
which would inspire future cooperative caching schemes.
The numerical results demonstrate up to 32% improvement in the outage
performance (and 16% on the average) for the studied plausible scenarios
where the probability of outage is below 0.5.
43 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
44. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Future Work
Our work in the first part could be oriented as follows:
Implement a distributed algorithm that makes use of the main characteristics of
OCC paradigm.
Analyzing on the effect of mobility patterns on the opportunistic caching
paradigm.
Extend the opportunistic content caching scheme considering the privacy and
anonymity issues.
Develop a distributed and cooperative algorithm to calculate the optimum
caching time for a specific data chunk in order to utilize the node’s storage.
44 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
45. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Future Work
Continue ..
The second part of this work can be extended along the following research
directions:
Generalize the model to relax few assumptions of practical relevance (N, γ,
Tc ).
Model and quantify the diversity gains attributed to nodes’ cooperation.
Quantify the outage performance for other vehicular mobility models.
Quantify the cooperation diversity gain that is above and beyond the mobility
gains explored here.
Develop novel cooperative caching schemes that capture the structured nature
of vehicular mobility.
45 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
46. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Publications
Osama Attia, Tamer ElBatt, "On the Role of Vehicular Mobility in Cooperative
Content Caching", accepted in IEEE WCNC 2012, Vehicular Workshop, April,
2012.
Osama Attia, Tamer ElBatt, "Opportunistic Content Caching in Wireless
Networks", under submission.
46 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks
47. Introduction
Conclusions
Opportunistic Content Caching
Future Work
Vehicular Mobility Role in Cooperative Content Caching
Publications
Conclusions and Future Work
Thank You!
Any Questions?
47 / 47 Osama Gamal M. Attia Content Caching Paradigms in Wireless Networks