You are on page 1of 10

1

Privacy Preservation Scheme for Intelligent Public


Transport System: An Emergent Intelligence
Approach
Suresh Chavhan, Member, IEEE

Abstract—Metropolitan transportation is dynamic and non- more probability of occurrence of third party involvement for
linear complex system. In this system there are possibilities observing and analyzing the communication entities. These
of altering, monitoring, forging, and accessing private, public problems can be avoided and solved by using privacy tech-
and resource information of depot staffs and communicating
agents by unauthorized agencies in a metropolitan area. Existing nique, that is, anonymity. The technique anonymity hides real
solutions for protecting privacy of depot staff and communicating identity of communicating entities and interacts anonymously
agents do not adapt to the dynamic occurrence of real time with neighbors. The anonymous communicating entities be-
events information and hence they are not sufficient for the comes malicious whenever they compromise completely [4],
randomness and non-linear complex systems. In this paper, we [7], [8] which is a serious issue because it is impossible to
propose privacy preservation scheme for intelligent transport
system’s depot in a metropolitan area. The scheme provides identify and remove them from the network. To solve this kind
privacy during the exchange of information of vehicle allocation, of issue we need of a strong privacy preservation technique
dispatch, revocation, financial, and maintenance. The absence of for the communicating entities and removing their privacy i.e.,
proposed privacy preservation scheme leads to anomalies like making non-anonymous. It is achieved only when disclosing
impersonation of genuinity, malicious, and greedy staff. We use the real identification information of malicious entities and
Emergent Intelligence (EI) technique to collect, analyse, share,
and to take dynamic decisions to protect the privacy of depot’s making them no more anonymous.
staff. The EI technique brings autonomy, flexibility, adaptiveness, Most of the schemes proposed for anonymity use a public
robustness, self-organization and evolution to take into account key infrastructure (PKI). The commonly used PKI based in-
the randomness and non-linearity complex systems behavior of frastructures are RSA and elliptical curve cryptosystem (ECC).
transportation system in a metropolitan area. The proposed The ECC based schemes have less computational cost and
scheme is implemented in Crypto++ package and its results
indicate that the scheme efficiently preserves transport depots smaller key size compared to RSA-based anonymity schemes
privacy in a metropolitan area. for providing same level of privacy protection, hence ECC is
more suitable than the RSA [7]–[11], [14], [16], [19], [20] for
Index Terms—Agent; Elliptic Curve Cryptography; Emergent
Intelligence; Metropolitan area; Privacy; Pseudonyms. intelligent transportation.
Group signature based schemes [3]–[5], [9], [10] and the
pseudonyms authentication schemes [3], [6]–[8], [11]–[15],
I. I NTRODUCTION [27] are recorded as anonymous authentication schemes to
The metropolitan areas are highly congested with more than address the security and privacy issues in ad-hoc networks.
one urban areas, satellite cities, rural areas, etc. It is divided In ad-hoc network all these schemes have computational
into regions and each region is again subdivided into zones. power and bandwidth limitations. PP-CP-ABE scheme [21]
In each region a depot is built (especially metropolitan cities preserves the privacy of users using access policies. These
in India) to meet the required needs. A depot is a premise access policies are formed using user attributes. Furthermore,
where resources (e.g., vehicles, fuel and staffs) are stored, authors have proposed PP-AB-BE scheme here they used
managed and allocated by the manager (e.g., agent). Due to hidden access policy, either with or without specifying the
the heterogeneous behavior of commuters’ arrival rates, staff, receivers and showed that it reduces the storage and communi-
and vehicles in transport depots [1], [2], there may be the cation overhead. A broadcast group key management (BGKM)
possibilities of altering, monitoring and forging the public and scheme [22] formalized for managing the key and proposed
private information of transport depots and staff as well by an attribute based access control policies while preserving the
third parties during interaction [3]. These third parties can user’s identity attributes privacy for sharing the documents in
use the private and public data to create traffic congestion, an untrusted cloud storage service. The approach is to encrypt
under utilization of surplus resources, blocking the lanes, etc. and decrypt the documents, whenever the document satisfies
Hence, the intelligent public transport system faces privacy the policies and keys, and is supported by BGKM scheme.
and security challenges [30]–[32] in a metropolitan city. Due Most of the above mentioned privacy preservation scheme
to the advancement of various information and communication faces some drawbacks like, computation overhead, disclosing
technologies (ICTs) [4]–[6] and their deployments resulted in anonymity of communication entities, identification and revo-
cation of malicious communication entities, RSUs are easily
Suresh Chavhan is with the Automotive Research Centre, School
of Electronics Engineering, VIT University, Vellore, Tamil Nadu, India. compromisable and vulnerable to physical attacks, they take
Email:suresh.chavhan046@gmail.com more latency for pseudonymous key generation, they will not
preserve and disclose privacy depending upon circumstances, which issues initial security keys and parameters. 2. RoadSide
and they are not efficient for distributed environments. Units (RSUs): They are installed in an open environment at
In this paper, we propose a privacy preservation scheme for the road side, intersections, etc. They are more possibility
Intelligent Public Transport System’s depots using Emergent of occurrence of physical breaches, therefore RSUs are not
Intelligence Technique in a metropolitan area. The proposed fully trusted. We assume that RTA monitors continuously the
privacy preservation scheme is based on the policies of trans- functions of RSUs and detects their malicious behavior in a
port depot’s staff, Pseudonmous technique, Cryptographic tech- limited periods of time.
niques, Bilinear pairing and Emergent Intelligence technique. Communication Network Model:
In the intelligent public transport system’s depot, staffs are The proposed privacy preservation scheme’s communication
divide into three levels depending upon their grades and at network model at a region in a metropolitan area is shown
each level different percentage of privacy will be provided. in Figure 1. Communication network model consists of agent,
The level of privacy will be determined by using policy, RSUs, vehicles and RTA. The regional office, RTA, contin-
which is formed and given by the regional trusted authority uously monitors and manages RSUs. The presence of RTA
at a depot. These policies are formed using credentials of helps to revocate malicious behavior of agents by sending
depot staff which contains type of staff, working time, work- the revocation developed technology, especially for vehicle to
ing place, authentication information, signature of regional vehicle (V2V) and vehicle to roadside unit (V2R) communi-
trusted authority, and pseudonyms. The Emergent Intelligence cation, called dedicated short-range communication (DSRC).
technique is used for collection, analysis, sharing and taking Time synchronization among RSUs and agents is assumed for
dynamic decisions like degree of information need to be validating the pseudonyms time to live.
shared among depot staff and revocation of depot staff in
a metropolitan area. The policy based preservation scheme
provides the accurate and reliable information (like resource
available, resource allocated, traffic conditions, etc.) to the
transport depot’s agents for sharing with the neighbor depot’s
agents.
The rest of the paper is organized as follows. Section II
describe definitions, assumptions and concepts required for
the proposed scheme. In Section III presents principle of EI
technique. In Section IV and V describe the principles of
privacy preservation scheme and policy based transport depot
staff. Section VI includes performance analysis of proposed Fig. 1. Communication network model at a region in a metropolitan area
schemes and results; and conclusions are drawn in Section
VII. Policy based Privacy:
A policy specifies the constraints under which a specific action
II. D EFINITIONS AND A SSUMPTIONS USED IN WORK can be performed on a certain sensitive resource [25]. There
are two cryptographic primitives for enabling the ’privacy-
In this section, we briefly describe definitions, assumptions,
aware’ policy enforcement:
communication network model, attack model, and some math-
ematical concepts background necessary for proposed privacy 1) policy based encryption: It allows to encrypt data ac-
preservation scheme in a metropolitan area. cording to a policy so that only entities fulfilling the
Public Transport Depot: policy are able to successfully perform the decryption
In metropolitan area a public transport depot is transport and retrieve the plain text data.
system’s operating base [23]. The depot provides maintenance, 2) policy based signature: It allows to generate a digital
servicing, monitoring and parking accommodation for vehi- signature on data with respect to a policy so that only
cles; Allocating and dispatching vehicles; and it also provides entities satisfying the policy are able to generate a valid
facilities, administrative functions, engineering and managerial signature.
functions for staffs. At depot there are three major grades of These cryptographic primitives have many applications, such
staffs, i.e., Depot Manager (DM); Operations and Engineering as access control, privacy policies, trust establishment, and
Manager (OEM); and Administrative, Personnel and Accounts automated trust negotiation. In this paper, we use policy based
Staff (APAS). These staffs are called as Depot Staff (DS) in encryption cryptographic primitive for privacy preservation of
a metropolitan area. Each of these staffs do their respective transport depot staffs during the exchange and/or access of
functions. Whenever there is a request from the neighbor depot information about the vehicle allocation, dispatch, revocation,
about the scarcity of resources (vehicles, staffs, etc.), then financial, and maintenance. At the transport depot without
public transport depot provides the required amount and type the privacy preservation scheme leads to the occurrence of
of resources needed. anomalies, like impersonation of a genuinity, malicious and
Assumptions: greedy staffs in a metropolitan area.
1. Regional Trusted Authority (RTA): It is deployed in the de- Attack Model:
pot. We assume that it is trusted and can not be compromised, Attack model is developed over the designed privacy policies.
During communication among entities (like vehicles, RSUs III. E MERGENT I NTELLIGENCE T ECHNIQUE (EIT)
and agents) in the metropolitan area there is a possibility The EIT is an extension of Multi-Agent System (MAS)
of occurrence of attackers. These attackers can be easily where agents group activities and individual independent
monitor, forge and alter more sensitive information, such as decisions making are incorporated. The EIT is collective
surplus resource, deficit resource, depot’s history, allocated intelligence of a group of agents [24], [35]–[37]. These
resource, etc. The malicious tasks are done by attackers group of agents cooperatively (whenever agents are compatible
in the metropolitan area. Here, we classified attackers as towards completing tasks execution), coordinately (whenever
internal and external. The external attacker observes ongoing there is need of agents joint actions for executing tasks) and
communication and analyzes traffic related data, but they can collaboratively (whenever agent has insufficient ability to carry
not decipher messages. The malicious agents and staffs are out tasks) interacts among themselves to provide dynamic
internal attackers, and they have full rights to access both independent decisions for the given problems. The EIT can be
public and private information of transport depots and hence used for independent tasks and their sub-tasks, which could be
they are very powerful attackers. There are several possible parallelly executes and provides partial (or complete) solution.
types of attack scenarios in metropolitan areas, some of them
are listed as follows:
1. Impersonate a genuine agent: The attacker may pretend to
be staff to fool others.
2. Malicious staff: The malicious behavior of staff may results
into illegal access of information (for example personal and
public). This may dangerously affects during emergency situ-
ations. The attackers may act like malicious vehicles, RSUs
and staff to carry out their functions.
3. Greedy staffs: The greedy behavior of staffs try to use Fig. 2. Scenario of 3 nodes network with task and its sub-tasks.
resources just for their own goals. The serous problem may
results during peak periods and traffic jam. This is because To illustrate consider the task tA in 3-nodes network (See
prediction condition may become false and therefore genuine Figure 2). As per the objectives of the task, it is possible to
staffs may affect their services with significant delay. split the task tA into 3-sub-tasks (stA1 , stA2 and stA3 ) which
Our scheme handles all the above mentioned attacks and can be independently solved by the 3 agents. Since the task is
ensures that privacy of transport depot staff and anonymity initiated at node A and the EIT is deployed in it, which creates
of agents communication are preserved, which are discussed 3-agents A1, A2 and A3 and migrates them to all 3-nodes A,
in the subsequent sections. B and C, respectively. The task tA is independently solved by
Background of mathematical concepts required for privacy: all the agents by considering the relevant information available
The basic mathematical concepts required for the design of locally and globally. Finally, the decision is communicated to
our proposed scheme are presented in this subsection. the main task solving agent A1 as per the following equation
1) Bilinear Pairing: The bilinear mapping is a function
D(tA ) = D(stA1 ) + D(stA2 ) + D(stA3 ) (1)
used in our proposed scheme, which uses pairing be-
tween elements of two groups to another group element where D(stA1 ), D(stA2 ) and D(stA3 ) are the partial or full
with a mapping function. The definition of bilinear decisions taken at nodes A, B and C, respectively. The same
mapping provides following properties [26]. problem would have been solved by the MAS exclusively at
Definition- Let two groups g1 and g2 with same order p, node A by collecting the relevant information from nodes B
where p = q n , q is a prime and n ∈ Z+ , g1 and g2 are and C.
two additive and multiplicative groups respectively, the In the literature, we can observe that some of the attempts
bilinear mapping ê : g1 × g1 → g2 satisfies following 3 made to use the EI for solving problems [33], [34]. But it is not
properties. clear, how this technique can be used for the problem solving
a) Bilinearity: ∀X, Y ∈ g1 , ∀c, d ∈ Z∗p : ê(cX, dY ) = in the uncertainty domains (like metropolitan area, where
ê(X, Y )cd , where Z∗p = [1, 2, ..., p − 1] dynamic demands, requirements, etc., occurs frequently).
b) Non-degeneracy: If X is a generator of g1 , then In this paper, transport depot’s staff information, such as
ê(X, X) is a generator of g2 . private, public and resource information collection, analysis,
c) Computability: There is an efficient algorithm to sharing and taking dynamic decisions to preserve privacy of
compute bilinear map ê : g1 × g2 them are done using EIT in a metropolitan area.
2) Elliptic Curve Discrete Logarithm Problem (ECDLP):
Given points X and Y in the group, find a number k IV. P OLICY BASED PRIVACY PRESERVATION SCHEME FOR

such that Xk = Y . The ECDLP is a well known hard INTELLIGENT TRANSPORT SYSTEM ’ S DEPOT STAFF
problem described in [?], [10], [11]. As we know the metropolitan area is divided into regions,
3) Bilinear Diffie-Hellman Problem (BDHP): Given and in each region a depot is built. In this section, we present
(X, a.X, b.X, c.X) for a, b, c ∈ Zp and it computes: policy-based privacy preservation system setup, three levels
e(X, X)abc [26]. of depot staffs privacy preservation at depot and Emergent
Intelligence technique based privacy preservation among depot provided by RTA using signature procedure. An assertion is
staffs. Table I shows the notations and their description used denoted as A ∈ {0, 1}∗. These credentials are generated using
in proposed scheme. the CredGen algorithm by RTA whenever an assertion is valid.
CredGen: The credential generation (CredGen) algorithm
TABLE I takes input as valid assertion A and RTA’s master key β, and
N OTATIONS USED IN TRANSPORT DEPOT STAFFS PRIVACY POLICY it gives output ξ(RT AP K , A) = β.h0 (A).
Notation Description
Here, we define a policy using logical expressions, which
k Secret key consists of conjunctions (∧) and disjunctions (∨) with creden-
β RTA’s master key tials of users. All these credentials are defined by the regional
C Policy encryption trusted authority (RTA) and it certifies the validity of them.
P oli ith policy The proposed policy-based privacy preservation scheme in a
C̄ Policy decryption depot has three levels of staffs depending upon grades of
M Message
them. The hierarchy of staffs in depots are as follows: (1)
Certi Certificate of ith level staff
Administrative, Personnel, and Accounts staff, (2) Operation
CIDi Certificate id of ith level staff
MID Manager id
and Engineering Manager, and (3) Depot Manager as shown
SigAlg Algorithm used to create to signature
in Figure 3. The proposed policy-based privacy preservation
RID RTA’s id
Val Validity of certificate
P KM ID Public key of MID
RT AP K RTA’s public key
RT ASK RTA’s private key
MT ype Message type either request or response

A. System Setup
The policy based privacy aware cryptosystem (PAC) setup
procedure is specified by two randomized algorithms, that
Fig. 3. Hierarchy of staffs at depot
is, (1) PAC setup, and (2) Regional Trusted Authority (RTA)
setup.
scheme provides different levels of privacy to each hierarchy
PAC setup: Given a secret key k as input parameter and do
of staffs at depot depending upon the privacy parameters (as
the following:
shown in Figure 4). The privacy preservation at level 1, 2, and
1) The BDH algorithm generates (p, g1 , g2 , ê) parameters 3 are describe as follows:
2) Randomly choose a generator X ∈ g1
3) n is random number chosen from N∗ , let M = {0, 1}n.
4) Let C = g1 × ({0, 1}n)∗ × M and S = (g2 )∗ × g1 .
5) Hash functions: h0 : {0, 1}∗ → g1 , h3 : {0, 1}∗ →
Z∗p , h4 : {0, 1}∗ → {0, 1}n, h5 : {0, 1}n →
{0, 1}n, and h6 : {0, 1}∗ → Z∗p
6) P ubP aram = (p, g1 , g2 , ê, n, X, h0 , h3 , h4 , h5 , h6 )
The public parameters specify different groups and public
functions that will be used in procedures and by participants
in the system.
RTA setup: The regional trusted authority (RTA) chooses Fig. 4. Privacy levels at depot
randomly a master-key s ∈ Z∗p , and which is used to compute
the corresponding public key RT AP K = sX. All system
participants know the public key.
C. Privacy preservation at level 1
B. Transport Depot Staffs Privacy Preservation Model The administrative, personnel, and accounts staff (APAS)
at depot get registered to the RTA. During registration the
In this subsection, we discuss 3-different privacy policies
APAS provides their private information to RTA, and the RTA
for each level of depot staffs for preserving their privacy at
hides these private information and stores in its database. The
transport depots in a metropolitan area.
RTA provides authentication information (Auth-Info), such as
We define an assertion to be declaration of staff at depot.
username/password. The RTA at a depot forms policy using
An assertion provides information about the staff’s attributes,
credentials of APAS, and is given as:
properties, capabilities and so on, which is encoded as binary
strings and representations of them is out of scope of this paper. pol =< Depot , x : Staf f > ∧ < Depot , x : ID > ∧ < Depot , x : T ime >
1 i i i
Assertions are represented as credentials and their validity is ∧ < Depoti , x : Location > ∧ < RT Aj , x : Auth − Inf o >
TABLE II
N OTATIONS AND THEIR MEANING USED IN PRIVACY POLICY EXECUTION

Notation Meaning
staf f ↼ ξ(RT AP K , A) ’Staff’ has been issued credential
ξ(RT AP K , A)
staf f ⇋< RT AP K , A > ’Staff’ fulfills condition
< RT AP K , A >
staf f ⇋ pol1 ’Staff’ satisfies the policy ’pol1 ’
staf f ⇋ pol1 ⇔ staf f ↼ Set of credentials fulfills policy ’pol1 ’
ξ1,2,...,j (pol1 )

where, x ∈ {Administrative staf f, P ersonnel staf f,


Accounts staf f } and ID is the identity of staff X. Fig. 5. Mutual authentication between level 1’s and 2’s staffs sequence
The following steps are used to preserve privacy of staffs diagram
using notations mentioned in the Table II depending upon
policy 1 issued by RTA:
sending the token, and is given as token =<
1) staf f ↼ ξ(RT AP K , A),
ESK {Staf fID ||N once||M T ype} >, where M T ype
2) The APAS encrypts message M according to pol1 , i.e.,
is message type and its value is either 1 (request or
C = P olEnc(M, pol1 ) (given in algorithm 1) and sends
permission message) or 2 (response message).
to RTA.
2) The Level 2’s manager presents its certificate, that
3) The RTA decrypts using its secret credentials, C̄ =
is, represented as Cert2 and is given as follows:
P olDec(C, pol1 , ξ1,2,...,j (pol1 ) (given in algorithm 2).
Cert2 =< S2 ||ESK {H(S2 )} >, where S2 =<
4) The RTA checks U = Ū , then staf f ⇋<
{CID2 ||M ID||SigAlg||Sig(RT ASK , M )||RID||V al||
RT AP K , A > and staf f ⇋ pol1 implies staf f ⇋
P KMID } >, where CID2 represents certificate id
pol1 ⇔ staf f ↼ ξ1,2,...,j (pol1 ).
of level 2, M ID is the manager id, SigAlg is the
5) The APAS performs its job, such as processing cash,
algorithm used to create signature, Sig(RT ASK , M ) is
vehicle information, staff information, etc.
the actual signature of RTA to verify genuinity of entity,
RID is the RTA id, V al is the validity of certificate,
Algorithm 1 Encryption based on policy 1
P KMID is public key of MID.
1: Data: message M and pol1 .
2: ti choose from {0, 1}n . 3) The level 1’s staff sends this certificate to the RTA for
3: Compute t = ⊕ai=1 ti , r = h3 (M k t k pol1 ), and U = rX. verification.
4: for i=1, Q
2, ..., a do
5: gi = k=1
ai
ê(Ri,k , ho (Ai,k )) 4) If the verification is successful then Level 1’s staff
6: vi = ti ⊕ h4 (gir k i) sends its certificate to the Level 2’s manager, that is,
7: end for
8: Calculates w = M ⊕ h5 (t) 
Cert1 =< S1 ||ESK {H(S1 )} >.
9: Ciphertext C = U, [vi ]1≤i≤a , w 5) The level 2’s manager sends this certificate to the RTA
10: End
for verification.
6) If successful, then the level 2’s manager grants permis-
Algorithm 2 Decryption based on policy 1 sion to send information.
1: Data: C, pol1 , ξ1,2,...j (pol1 ).
2: for i=1,2, ..., a do Pai  D. Privacy preservation at level 2
3: Compute g¯i = ê U, k=1 (ξi,k , Ai,k )
4: ¯
ti = vi ⊕ h4 (ḡi k i) The operation and engineering manager (OEM) at depot get
5: end for
6: Compute M̄ = w ⊕ h5 (⊕a ¯
i=1 ti ).
registered to RTA. During registration the OEM provides their
7: Compute Ū = h3 (M̄ k ⊕a ¯
i=1 ti k pol1 ).X private information to RTA and the RTA hides these private
8: End
information and stores in its database. The RTA provides au-
thentication information (username/password) and confidential
Before the level 1’s staff starts sharing information, such information (signature from the RTA). Using credentials of the
as processed cash, vehicle information, staff information, OEM forms following policy, pol2 , by the RTA at depot and
etc., with the level 2’s manager, they have to authenticate is given as:
each other, i.e., mutual authentication is required. We use
certificate based authentication scheme, because the certificate pol2 =< Depoti , y : Manager > ∧ < Depoti , y : ID > ∧ < Depoti , y : T ime >
is issued by regional trusted authority and is more secure than ∧ < Depoti , y : Location > ∧ < RT Aj , y : Auth − Inf o > ∧
other scheme and is shown in Figure 5. The actions taken < RT Aj , y : Sig(β, M) >
during the mutual authentication are as follows: where y ∈ {Operation manager, Engineering manager},
ID represents manager identity information, β is private key
1) Level 1’s staff asks permission to send informa- of the RTA, and M represents message.
tion, such as processed cash, vehicle information, For preserving privacy of depot staffs at level 2 using pol2 ,
staff information, etc., from Level 2’s manager by we have to repeat the same steps as described in the IV-C
and in those steps we need to replace the privacy preserving
policy from pol1 to pol2 with having same parameters. Before
level 2’s manager starts sharing information, such as allocated
buses and crews, dispatched, maintenance, etc., with level 3’s
depot manager, they have to mutually authenticate each other
using the procedure described in the IV-C for level 1 staffs
by replacing the level 1 to level 2, level 2 to level 3 and
pol1 to pol2 in the sequence diagram.

E. Privacy preservation at level 3


The depot manager (DM), is an agent, at the depot get
registered to RTA. During registration the DM provides its
private information to RTA and the RTA hides these private
information and stores in its database. The RTA provides
authentication information (username/password), confidential
information (signature from RTA), and pseudonym. Using
Fig. 6. Interaction among intelligent transport system’s depot staff
credentials of the DM forms policy by the RTA at a depot
and is given as:
(5) During these interactions, they shares and accesses cer-
pol3 =< Depoti , z : DM > ∧ < Depoti , z : T ime > ∧ < Depoti , z : Location >
tain degree of private and/or public information of same
∧ < RT Aj , z : Auth − Inf o > ∧ < RT Aj , z : Sig(β, M) >
∧ < RT Aj , z : P seudonym >
or other level of staffs and depots.
(6) Finally, the EIT at initiator depot takes decision of an
where pseudonym is an unreal identity given to the depot accurate percentage of type of information need to be
manager or agent by regional trusted authority using their shared with particular depots staffs and also revocation
credential data. of malicious depot staffs from the depot.
For preserving privacy of depot staff at level 3 using pol3 , The static agent (SA) deployed at each depot, which takes
we have to repeat the same steps as described in the IV-C and care of depot manager (DM) functions in a metropolitan area.
in those steps we need to replace the privacy preserving policy The SA creates and dispatches mobile agents (MA) to depot
from pol1 to pol3 . The depot manager has the rights to access staffs level 1 and 2 in a depot as shown in Figure 6. These
to level 1’s and level 2’s information. migrated MAs collect and analyze public data of depot and
The privacy preserved depot staff’s public and private private data of depot staffs, and shares with SA. The SA
information are exchanged in the subsequent subsection. analyzes collected information from group of depot staffs and
takes dynamic decisions to achieve a common goal. During
F. Transport depot staff’s public/private data collection and taking decisions there may be the possibility of occurrence of
sharing change of depot staff’s levels and emergency incidents. These
sudden changes are incorporated in the privacy preservation
In this subsection, we describe depots public data (contains
scheme, hence EI technique is more adaptable. During adap-
resource availability, resource allocated, resource dispatched,
tation it defines scaling factors to disclose certain percentage
parking information, maintenance information, etc.) and staffs
of information depending upon the depot staffs level and type
private data (contains identity, habits, method of driving, sig-
of emergency incidents.
nature, etc.) collection, analysis, sharing and taking dynamic
EI technique formed group consists of mutually authenti-
decisions, like degree of information need to be shared among
cated depot staffs. These depot staff shares information with
depot staffs in order to achieve a common goal by using
some constraints, and these constraints are defined by the EI
Emergent Intelligence technique in a metropolitan area.
technique depending upon the change of depot staffs level,
Steps of decision making Emergent Intelligence technique at
sudden occurrence of emergency events, and so on. These
depot in a metropolitan area:
constraints are used to define the value of scaling parameters,
(1) Initiator depot’s agent uses staff, RSUs, agents and such as α, β, γ, λ, µ, and ζ, they take values in the range
vehicles to form a cluster or group. of [0, 1]. Three important values of these parameters are
(2) Analyzes and estimates resource required and decides described in Table III: The EI technique uses probabilistic
the type of information need to be shared among them, model to define the amount of privacy need to be disclosed and
also waiting time required at each depot. preserved among the depot staffs level. Let i indicates APAS, j
(3) During emergency situations, it uses historical and an- indicates OEM, and k indicates DM. The probabilistic privacy
alyzed information for anticipating resources needed, model, which defines the amount of privacy preservation and
traffic jams, waiting time, etc. disclose of private and public data of depot staffs level, is
(4) These anticipated information are used for deciding shown in Figure 7.
staffs and agents cooperation, competence and negoti- The depot staffs of level i’s probability of disclosing private
ation interaction. and public (shareable) data to j and k are denoted as Pdi,j and
TABLE III The misbehavior of legitimate staff revocation in a metropoli-
L EVEL OF PRIVACY PRESERVATION OF TRANSPORT DEPOT STAFFS tan area at depot is required to avoid insider misbehavior
attacks from revealing the confidential information. The mali-
Scaling Description Level of privacy preservation cious behavior of legitimate staff is observed by upper, lower,
param-
eters
and same level staffs at depot in a metropolitan area during
value interactions with them. The observed staff identifies misbehav-
1 Transport depot staffs completely dis- 0 (privacy is disclosed) ior of legitimate staff using their policies. The observed staff
closes both private and public informa-
tion records misbehavior events and creates misbehavior of legiti-
0 Transport depot staffs do not discloses 1 (completely privacy is pre- mate staff report (MSR), i.e., M SR =< M IS, P oli , IDx >,
both private and public information served)
where M IS is the misbehavior information, P oli is the policy
(0,1) Transport depot staffs partially dis- (0,1) (partially privacy is pre-
closes both private and public informa- served and/or disclosed) of ith level, and IDx is the identity of misbehavior legitimate
tion staff x. All observed staff sends the M SR, after encryption, to
the RTA. The RTA decrypts message and takes average of all
the reports taken from the observed staffs, and it takes average
feedbacks from all and uses IDx and P oli to take a decision
of revocation of staff x.

V. P ERFORMANCE A NALYSIS
In this section, we analyze performance of the system by
considering the performance measures, such as latency of
schemes, revocation latency, response time and execution time.
The proposed scheme has been scripted and implemented in
Fig. 7. Privacy model of depot staff
C++, we used the Pairing-Based Cryptography (PBC) Libraray
[29] for elliptical curve and pairing operations and we also
i,k used the Cryptopp package [28] for implementation of pro-
Pd respectively, and are given as
posed schemes. The implementation has been carried out on a
Pdi,j = P oli × P olj × Psi,j × αPpi,j and Pdi,k = P oli × P olk × Psi,k × ζPpi,k
(2) dual-CPU Intel Core i5-2400 at 3.10 GHz desktop computer
with 12-GB RAM running Fedora 25. The performance anal-
where Ps is public or shareable data and Pp is private data ysis results were averaged over 500 randomized simulation
of staff i. Similarly, probability of not disclosing private and
public data are given as follows: runs. We chose key sizes of ECC as 224 and RSA as 2048,
i,j
and ECC as 2048 and RSA as 3078 bits for providing same
Pnd = 1 − Pdi,j and Pnd
i,k
= 1 − Pdi,k (3)
level of security.

The probability of disclosing private and public or shareable


data of operational engineer and manager level j to i and j to 12
ECPP

k are denoted as Pdj,i and Pdj,k respectively, and are given as


Time taken for pseudonyms generation (in msec)

Proposed Scheme
10 RSA

Pdj,i = P olj × P oli × βPsj,i and Pdj,k = P olj × P olk × Psj,k × µPpj,k 8

(4)
6

Similarly, probability of not disclosing private and public data 4


are given as follows:
j,i 2
Pnd = 1 − Pdj,i and Pnd
j,k
= 1 − Pdj,k (5)
0
The probability of disclosing private and public or shareable 2 4 6 8 10
Number of agents
12 14 16

data from depot staff, depot manager, level k to i and k to j


are denoted as Pdk,i and Pdk,j respectively, and are given as Fig. 8. Time taken for pseudonyms generation transport depot manager
k,i k,i k,j k,j
Pd = P olk × P oli × γPs and Pd = P olk × P olj × λPs (6)
Figure 8 shows the comparative results of ECPP [18], RSA
Similarly, the probability of not disclosing private and public and Proposed Scheme. The Proposed Scheme takes less time
data are given as follows:
for generation pseudonyms for transport depot manager as
k,i
Pnd = 1 − Pdk,i and Pnd
k,j
= 1 − Pdk,j (7) compare to ECPP. RSA performs better than the proposed
scheme, but it does not provide the level privacy provided
by the Proposed scheme.
Figure 9 shows the time taken for revocating the malicious
G. Transport Depot Staffs Revocation transport depot staff in the metropolitan area. Let TAIM is the
Misbehavior of legitimate staffs (i.e., insiders) at depots in time required to identify malicious staff by neighbor staff S,
a metropolitan area are more difficult and complex to prevent. TSMSR is the time required to prepare M SR by S, TSE is
D
Because insiders possess credentials and policies issued by the the time required to encrypt the M SR by S, TRSU is the
ID
RTA to perform authentication and their respective functions. time required to decrypt the M SR by RSU, TRT A is the
350 TABLE IV
Proposed Scheme
ECPP Scheme
N OTATIONS AND E STIMATED EXECUTION TIME FOR CRYPTOGRAPHIC
300
OPERATIONS
Time taken for staff revocation (in msec)
250

200
Notation Description Execution
150
Time
Tmul Time for 1-point multiplication in g1 0.6 msec
100
Tpair Time for one pairing operation 4.5 msec
50 Txor Time for XOR operation 0.7 msec
0
Tcanct Time for concatenation operation 0.1 msec
0 2 4 6 8 10 12 14 16
Number of staff Tcred Time required for issuing credentials 3.5 msec
Tverf Time required to verify the policy with staff 6.5 msec
Fig. 9. Time taken to revocate the transport depot staff

TABLE V
RL E STIMATED EXECUTION TIME FOR DIFFERENT POLICIES FOR PRESERVING
time required to obtain the identity from RT A, TRT A is the PRIVACY OF STAFFS AT DEPOT
transmission time for the revocation list by RT A, T T T L is
the amount of remaining time of pseudonym to expire. The
Notation Description Execution
revocation time is denoted by T RL is given by Time
pol
Tpriv1 Execution time of preserving the privacy of the 52 msec
T RL = TSIM + TSM SR + TSE + TRSU
D ID
+ TRT RL
A + TRT A + T
TTL staff using policy 1
pol
Tpriv2 Execution time of preserving the privacy of the 62 msec
From literature, the ECPP scheme is the only scheme used to pol
staff using policy 2
revoke nodes and we compare our proposed scheme with it. In Tpriv3 Execution time of preserving the privacy of the 75 msec
staff using policy 3
our proposed scheme, searching and revoking the agent takes
very less time by the RTA and roadside units in the metropoli-
tan area compare to ECPP scheme (because it takes more
number of multiplications and pairing operations). Hence, our shows the execution time of different cryptographic operations.
pol1
proposed scheme is faster. Similarly, revocation of transport Let Tpriv is the execution time for verifying the privacy
pol1
depot staffs is done. preservation based on policy 1, Tcred is the time required to
pol1
issue credentials of staff, Tenc is the time required to encrypt
pol1
5000
the message based on policy 1, Tdec is the time required
pol1
4500
ECC 256
RSA 3072 to decrypt the message based on policy 1, Tver is the time
ECC 224
4000 RSA 2048 required to verify the policy 1. Table IV shows the different
Response Time (in milliseconds)

3500
cryptographic operations required during the policy execution.
3000
pol1
2500
The Tpriv is given as follows:
2000
pol pol pol pol pol
1500
Tpriv1 = Tcred1 + Tenc1 + Tdec 1 + Tver 1
1000
where,
500
pol1
0
0 10 20 30 40 50 60 70 80 90 100
Tenc = (TXOR + Tcant + Tmul + a(ai Tpair + 2TXOR + Tcant ))
Pseudonymity

pol1 2

Fig. 10. Response time with varying number of pseudonyms Tdec = Tcant + a(Tpair + TXOR + TXoR + Tcant + TXOR Tcant )

Figure 10 shows the response time of proposed scheme pol2 pol3


Similarly, for policy 2 and 3 are Tpriv and Tpriv respectively
with varying number of pseudonyms. For minimizing the are given as follows:
communication delay, we use the Public Key Cryptography pol pol pol2 pol pol2
Tpriv2 = Tcred2 + Tenc + Tdec 2 + Tver
(PKC), i.e., Elliptic Curve Cryptography (ECC) and compared
with the RSA cryptography. Because of the lower computa- pol pol pol3 pol pol3
Tpriv3 = Tcred3 + Tenc + Tdec 3 + Tver
tional cost and smaller key size of ECC results into more
response time compared to RSA as shown in Figure 10. The table V shows execution time of policy 1, 2, and 3 for
Even though the key size of the ECC is smaller compare to preserving the privacy of the staffs at depot in a metropoli-
RSA, it provides the equal amount of security strength. The tan area. The policy 3 has more execution time than other
ECC-based authentication generates lesser packets compared because of pseudonyms of Depot Manager i.e., agent and it is
to RSA-based authentication (which generates much larger periodically changes its value and are shown in Figure 11.
packets). Figure 10 shows the different sizes of the PKC We estimated the probability of privacy loss of levels 1 to 3
key and their response time with the different number of with different percentage of private and public data disclosed
pseudonymity of depot manager in the metropolitan area at at depot in a metropolitan area is shown in Table VI.
a depot. The Figure 12 shows the probability of privacy loss of
Execution time is the time spent for verifying the privacy different levels of depot staffs with varying amount of private
preservation of staffs based on policy 1, 2, and 3. Table III and public data disclosed at depot in a metropolitan area. The
80
1
Private data
70
0.9 Public data

60 Level 1
0.8

Execution time (in msec)


50
0.7

Probability of privacy loss


Level 2
40
0.6
30 Level 3
0.5
20
0.4
10
0.3
0
Policy 1 Policy 2 Policy 3 0.2
Privacy preservation policies

0.1

0
Fig. 11. Execution time of preserving the privacy of the staffs using policy 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
Percentage of private and public data disclosed
0.9 1

1,2. and 3.

Fig. 12. Probability of privacy loss of levels 1, 2, and 3.


TABLE VI
E STIMATION OF PROBABILITY OF PRIVACY LOSS OF LEVELS 1 TO 3 WITH
PERCENTAGE OF PRIVATE DATA DISCLOSED (PP DD ) AND PUBLIC DATA
DISCLOSED (PP uDD )

Number of Transport Depot Staff Accessing Data


15
Level PP DD PP uDD
1 0.06 1.0 10

2 0.04 0.8
3 0.01 0.4 5

0
12
10
25
8 20
6 15
4
proposed scheme preserves the privacy of the private data of 2
Percentage of Private and Public Data disclosed 0 0
5
10

Number of Intelligent Transport System’s Depot Staff


depot staffs and it is flexible with the public data. The privacy
is disclosed depending upon the depot staffs levels and the type
Fig. 13. Percentage of private and public data accessing by transport depot
of data, i.e., private or public data and is shown in Figure 12. staff
Figure 13 shows the percentage of staff’s private and public
data disclosed and protected from other staff.
[7] Sun, Yipin, et al. “An efficient pseudonymous authentication scheme
with strong privacy preservation for vehicular communications.” vehicular
VI. C ONCLUSIONS Technology, IEEE Transactions on 59.7 (2010): 3589-3603.
In this paper, we have proposed a novel privacy preservation [8] Li, Jie, Huang Lu, and Mohsen Guizani. “ACPN: a novel authentication
framework with conditional privacy-preservation and non-repudiation for
scheme for intelligent public transport system’s depots staff in VANETs.” Parallel and Distributed Systems, IEEE Transactions on 26.4
a metropolitan area. The proposed privacy preservation scheme (2015): 938-948.
provides privacy to depot staff depending upon policy given [9] Wasef, Albert, and Xuemin Sherman Shen. “PPGCV: Privacy preserving
group communications protocol for vehicular ad hoc networks.” Commu-
by the regional trusted authority in a metropolitan area. These nications, 2008. ICC’08. IEEE International Conference on. IEEE, 2008.
policies are formed by the regional trusted authority using [10] Blake, Ian F., Gadiel Seroussi, and Nigel P. Smart. “Advances in elliptic
credentials given by depot staff. The novel proposed scheme curve cryptography.” Vol. 317. Cambridge University Press, 2005.
protect the public and private data of transport depot and [11] Koblitz, Neal. “Elliptic curve cryptosystems.” Mathematics of computa-
tion 48.177 (1987): 203-209.
staff. Hence, the proposed privacy preservation scheme avoids [12] Lin X, Sun X, Ho PH, Shen X. “GSIS: a secure and privacy-preserving
lane blockage, accidents, traffic jams in the metropolitan area. protocol for vehicular communications.” Vehicular Technology, IEEE
Performance analysis of the proposed scheme justifies privacy Transactions on. 2007 Nov;56(6):3442-56.
[13] Wasef A, Shen XS. PPGCV: Privacy preserving group communica-
preservation effectiveness. tions protocol for vehicular ad hoc networks. In Communications, 2008.
ICC’08. IEEE International Conference on 2008 May 19 (pp. 1458-1463).
IEEE.
R EFERENCES [14] Blake IF, Murty VK, Xu G. “Refinements of Miller’s algorithm for
[1] Du, Rong, et al. “Effective urban traffic monitoring by vehicular sensor computing the Weil/Tate pairing.” Journal of Algorithms. 2006 Feb
networks” Vehicular Technology, IEEE Transactions on 64.1 (2015): 273- 28;58(2):134-49.
286. [15] Raya, Maxim, Panos Papadimitratos, and Jean-Pierre Hubaux. “Securing
[2] Zhang, Xin, and Gang-len Chang. A Transit-Based Evacuation Model for vehicular communications.” IEEE Wireless Communications Magazine,
Metropolitan Areas, Journal of Public Transportation 17.3 (2014): 9. Special Issue on Inter-Vehicular Communications 13.LCA-ARTICLE-
[3] Qu F, Wu Z, Wang F, Cho W. “A security and privacy review of 2006-015 (2006): 8-15.
VANETs.” Intelligent Transportation Systems, IEEE Transactions on. [16] Zwierko, Aneta, and Zbigniew Kotulski. ”Mobile agents: preserving pri-
2015 Dec;16(6):2985-96. vacy and anonymity.“ Intelligent Media Technology for Communicative
[4] Lu, Rongxing, et al. “ECPP: Efficient conditional privacy preservation Intelligence. Springer Berlin Heidelberg, 2005. 246-258.
protocol for secure vehicular communications.” INFOCOM 2008. The [17] Cisse R, Albayrak S. An agent-based approach for privacy-preserving
27th Conference on Computer Communications. IEEE 2008 Apr 13. recommender systems. In Proceedings of the 6th international joint
[5] Papadimitratos P, Kung A, Hubaux JP, Kargl F. “Privacy and identity conference on Autonomous agents and multiagent systems 2007 May
management for vehicular communication systems: a position paper.” 14 (p. 182). ACM.
InWorkshop on standards for privacy in user-centric identity management [18] Lo NW, Tsai JL. “An Efficient Conditional Privacy-Preserving Authen-
2006. tication Scheme for Vehicular Sensor Networks Without Pairings.”IEEE
[6] Hubaux, Jean-Pierre, Srdjan Capkun, and Jun Luo. “The security and Transaction On Intelligent transportation systems, Vol. 17, No. 5, May
privacy of smart vehicles.” IEEE Security & Privacy Magazine 2.LCA- 2016.
ARTICLE-2004-007 (2004): 49-55. [19] Sun J, Zhang C, Zhang Y, Fang Y. “An identity-based security system
for user privacy in vehicular ad hoc networks.” Parallel and Distributed
Systems, IEEE Transactions on. 2010 Sep;21(9):1227-39.
[20] Zhou Y, Mo Z, Xiao Q, Chen S, Yin Y. “Privacy-Preserving Trans-
portation Traffic Measurement in Intelligent Cyber-Physical Road Sys-
tems.”IEEE Transaction on Vehicular Technology, 2015.
[21] Zhou Z, Huang D, Wang Z. “Efficient privacy-preserving ciphertext-
policy attribute based-encryption and broadcast encryption.” Computers,
IEEE Transactions on. 2015 Jan;64(1):126-38.
[22] Nabeel M, Shang N, Bertino E.“Privacy preserving policy-based content
sharing in public clouds.” Knowledge and Data Engineering, IEEE
Transactions on. 2013 Nov;25(11):2602-14.
[23] Information about Bus Depot, http://www.ppiaf.org/sites/ppiaf.org/files/
documents/toolkits/UrbanBusToolkit/assets/3/3.1/35(vii)b.html, accessed
on 2016 Jan 25.
[24] Chavhan S, Venkataram P. Emergent Intelligence Based QoS Routing
in MANET. Procedia Computer Science. 2015 Dec 31;52:659-64.
[25] Yung, Andrew S. Patrick Moti. “Financial Cryptography and Data
Security“. Springer publication, 2005.
[26] Martin, Luther. ”Introduction to identity-based encryption.“ Artech
house, 2008.
[27] Manvi SS, Tangade S. ”A survey on authentication schemes in VANETs
for secured communication“. Vehicular Communications. 2017 Mar 2.
[28] Wei Dai, ”cryptopp library,“ [Online]. Available:
https://www.cryptopp.com/wiki/Linux
[29] B. Lynn, ”The pairing-based cryptography library,“ [Online]. Available:
http://crypto.stanford.edu/pbc/
[30] Hasrouny H, Samhat AE, Bassil C, Laouiti A. ”VANet security chal-
lenges and solutions: A survey.“ Vehicular Communications. 2017 Jan
27.
[31] Karanki SS, Khan MS. ”SMMV: Secure multimedia delivery in vehicles
using rode side infrastructure.“ Vehicular Communications. 2016 Dec 23.
[32] Zarei M, Rahmani AM. ”Analysis of vehicular mobility in a dynamic
free-flow highway.“ Vehicular Communications. 2016 Dec 12.
[33] W. D. Hillis, ’Intelligence as an emergent behavior; or, the songs of
eden’, Daedalus (1988) pp. 175–189.
[34] T. Wolf, ’Analyzing and engineering self-organizing emergent applica-
tions’, 2007.
[35] Chavhan, S. and Venkataram, P., 2018. Commuters’ traffic pattern and
prediction analysis in a metropolitan area. Journal on Vehicle Routing
Algorithms, 1(1), pp.33-46.
[36] Chavhan, S. and Venkataram, P. (2019). Emergent Intelligence: A Novel
Computational Intelligence Technique to Solve Problems.In Proceedings
of the 11th International Conference on Agents and Artificial Intelligence
- Volume 1: ICAART, pages 93-102.
[37] Chavhan S, Venkataram P. Emergent intelligence technique-based trans-
port depot resource management in a metropolitan area. Journal on
Vehicle Routing Algorithms. 2019, Vol.no. 3, pp.:1-8.

You might also like