This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2018.2883629, IEEE
Communications Letters
IEEE COMMUNICATIONS LETTERS 1

An Energy-Efficient Transaction Model for the

Blockchain-enabled Internet of Vehicles (IoV)
Vishal Sharma, Member, IEEE,
1HWZRUN%ORFNFKDLQ 'LVWULEXWHG%ORFNFKDLQ
Abstract—The blockchain is a safe, reliable and innovative 2SHUDWLRQV 2SHUDWLRQV
,R9
mechanism for managing numerous vehicles seeking connectivity.
$
However, following the principles of the blockchain, the number
of transactions required to update ledgers pose serious issues for 0 0 0 0 0 0 0 0 0 0 0 0 00 00 00 00 00 00 00 00 00 00
000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 000 00
0 0 0 0 0 00 00 00 00 00 00 00 00 00 00 0 0 0 0 0 0
%6 % $ %

vehicles as these may consume the maximum available energy. To ,QWHUQHW $3V

&RUH
resolve this, an efficient model is presented in this letter which 9LUWXDOL]DWLRQ
%%83RRO
&

'
is capable of handling the energy demands of the blockchain- %6 '
&

,QWHUDFWLRQEHWZHHQFOXVWHUV
enabled Internet of Vehicles (IoV) by optimally controlling the (QHUJ\&RQVHUYLQJ3RLQWV
EDVHGRQVLPLODU
EORFNFKDLQDSSOLFDWLRQV
number of transactions through distributed clustering. Numerical (QHUJ\(IILFLHQW7UDQVDFWLRQV

results suggest that the proposed approach is 40.16% better in

terms of energy conservation and 82.06% better in terms of the Fig. 1. An exemplary illustration of IoV with energy conserving points.
number of transactions required to share the entire blockchain-
data compared with the traditional blockchain.
Index Terms—Blockchain, IoV, IoT, Energy Efficiency. derivatives for reducing the burden of transactions on each
device in IoV by finding the optimal slots for updating the
I. I NTRODUCTION blockchain ledgers. In general, the presented optimal transac-
Blockchain enables shared access to information which tion model is driven by the selection of appropriate Cluster-
is broadcasted across a network based on the trust of its Heads (CHs) as a part of distributed clustering. The success
participants [1]. Industries aiming at services for the Internet of the approach depends on the reduction in the overall cost
of Vehicles (IoV) consider blockchain as a leading technology of operation, which is expressed in terms of energy-efficiency
for handling managerial as well as transmission aspects of and the number of iterations performed to identify the final
vehicles [2]. Research groups in Mobility Open Blockchain transaction in the formulated blockchain.
Initiative (MOBI) have highlighted the potential use of this To further clarify, the problem at hand can be observed from
technology for IoV [3]. Challenges like broadcast collision- Fig. 1 in which each entity plays the role of a miner where
avoidance, resource scheduling [4], and privacy-preserving [5] consensus is performed through broadcasting. Let I denote
in IoV can be resolved through an efficient implementation of the set of vehicles each operating a set S of applications that
vehicular-blockchain. Blockchain enables IoV to be protected require blockchain procedures to accomplish their operations,
against different types of cyber threats as well as allows such that the total energy consumed by ith vehicle for |S|
secure distribution of vehicular services. In spite of playing number of blockchains is expressed as:
a pivotal role, there are certain issues related to the prominent |S|   
(t) (t) (t) (t)
X
use of this blockchain technology, which includes excessive BS,i = βC,M + βR + βU , (1)
i
i=1
energy consumption for ledger-updates, peer-to-peer and peer-
(t)
to-multi-peer smart contracts, and excessive transactions [6]. where βC,M is the energy consumed for security operations
Resolving these can help to extend the utility as well as the such as light-weight elliptic curve cryptography for checking
applicability of blockchain to IoV. (t) (t)
correctness of nodes, βR and βU are the energy requirements
In the networks where communications are bounded by of transmission procedures and blockchain-update (Ledgers)
blockchain technology, the number of transactions on each operations, respectively. In the proposed model,
node is considerably high and these transactions cause a huge (t)
βU = H (RC × ER ) , (2)
impact on the consumption of energy resources. Moreover, the
mechanisms in blockchain use public key operations, which and  
k
may cause considerable overheads even if light-weight cryp- (t)
X
βR = H (EC × γ)j  , (3)
tography algorithms are used for onboard security. However, j=1
the primary cause of concern is the multiple exchanges of
transaction-messages between the vehicles and the core, which where H is the number of intermediate hops, RC is the
is a problem and it is desired to find a solution to reduce these number of records updated in a transaction, ER is the per
number of transactions. record energy consumption, EC is the per request energy
consumption, γ is the number of requests and k is the types
II. S YSTEM M ODEL AND P ROBLEM S TATEMENT of messages like send, receive and acknowledgement. The
The solution presented in this letter uses distributed cluster- network conditions for optimality are based on governing
ing mechanism based on stochastic volatility model of security equations which are derived through a stochastic volatility
model as it helps to predict the variation in energy consump-
V. Sharma is with the Department of Information Security Engineer-
ing, Soonchunhyang University, Asan-si 31538, Republic of Korea, Email: tion of the network. It also allows to accurately calculate the
vishal sharma2012@hotmail.com. network sustaining rate if no additional batteries/resources are

1089-7798 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2018.2883629, IEEE
Communications Letters
IEEE COMMUNICATIONS LETTERS 2

provided to IoV, such that the network differential variation, major factors in the proposed approach are about the selection
by using Heston Model [7], is given as: of CHs, decision to transmit (when and how), the number
dB 
(t) (o)
 √ dB0 of permissible blockchain queries, and location-based ledger-
= λ BS,i − BS,i +  σ , (4) offloading. Satisfaction of all these issues through stochastic
dt dt
(o) volatility helps to sustain the network for longer durations.
where BS,i is the initial values for energy consumption, λ The mechanism of distributed clustering can be followed from
is the number of blockchain requests per unit time,  is the
ratio of the excessive energy to the total initial energy of the

Components
A A
vehicle, σ is the standard deviation of energy required w.r.t.

Core
B B
0
energy utilized, and dB D Global

dt denotes the rate of change of requests

CH selection
C D IoV to allow F Local Authentication
necessary Authentication
per unit time. Based on the given model, the location of a E F connections E
Removing frequent updates with
vehicle can be marked as L = f (x, y), and the maximum Removal of nodes to conserve transmissions.
In this case, "C" is prohibited from transmission core components. Distributing the
as it already belongs to CH-D. CHs for load balancing and energy-
distance to connect can be written as R, which shows the conservation

range between entities involved in the update and sharing of

Fig. 2. An exemplary illustration of the proposed mechanism of distributed
ledgers. To sustain at a given energy rate, the network can be clustering for the blockchain-enabled IoV.
modeled into a transfer function Ft , which is given as:
|C| |I| |S|   Fig. 2, according to which, every chain is labeled sequentially
(τ ) (A)
X XX
Ft = Pc × P(f (p)) , (5) to identify the nodes involved in updates. The regular oper-
i=1 j=1 k=1
ations and authentication procedures are the same as defined
(A) in the original blockchain. However, in the proposed model,
where Pcis the probability denoting the presence of a re-
ceiver, P(f (p)) denotes the probability function that a vehicle local and global chains are used for conserving energy, in
is in the range, C is the set of clusters, τ is the operational which the ones with critical energy values (marked in red) are
time, and removed from the global operations and an appropriate CH
R
is selected to handle their operations. For this, the local and
Z
P(f (p)) = 1 − Pc × f (p)dp, (6)
0 global authentication processes are used, which involves hash-
where f (p) is the vehicle movement function within R and Pc based authentication of CHs with the core components and
is the probability of existence of vehicles with the given move- hash-based authentication of general vehicles with CH. The
ment function (f (p)). Now, based on the above conditions, the CHs are temporary and their choices are made on-demand
network problem deals with attaining maximum transfers at a and as per the state of the network. This mechanism not
low consumption of energy, i.e., only reduces the burden of excessive sharing but also lowers

(τ )
 the complexity involved in performing fork operations for
max Ft , ∀ C, ∀I, ∀S, (7) individual entries. Following Lemmas help to understand the
s.t.  
operational details of the proposed methodology.
dB Lemma-1 In the given model, the sustaining rate of the
min , ∀ C, ∀I, ∀S,
dt
network is governed by f (p) and λ, and the capacity of the
τ0 0 network decreases if λ increases to a large extent. In such a
γ≤ , τ ≤ τ,
τ
case, if the rate of change in the number of requests approaches
|{z}
vehicle stay time
Z R zero, the lifetime becomes only a function of initial value
1− Pc × f (p)dp ≥ Pth ,
0 |{z} of energy and per-unit-energy consumption of the blockchain
threshold Probability
  application.
τ R
Proof: A blockchain network should limit to share the
Z Z
min  P(f (p)). (t) dp dt ∀ C, ∀I, ∀S. (8)
0 0 |{z}
error in transfer
operations for maintaining network secrecy. However, the
success of the proposed approach depends on the knowl-
III. D ISTRIBUTED CLUSTERING FOR edge of the operations. According to which, the network
B LOCKCHAIN - ENABLED I OV governing equations can be calculated based on two fre-
Usually, clustering involves a centralized entity, which quencies, f1 and f2 , of vehicle slots for operations defined
is against the principles of the blockchain. However, with in (1). Considering that the vehicles operate in a Gaus-
a distributed phenomenon over the selection of CHs, the sian mode, the two frequencies for network and blockchain
 
λ1 −λ1 2
−1
principles of blockchain remain intact and the network can operations can be given as f1 = 1
√ e 2 σ1
and
σ1 2π
be managed efficiently. Distributed clustering helps to lower −1

λ2 −λ2
2
1
the number of updates for ledgers as well as reduces the f2 = σ √ e , respectively. Gaussian distribution
2 σ2
2 2π
number of blockchain-links (ψ) generated for each ledger suites these formations as the average mean for network is
operation. Moreover, the slot-wise transactions offer better high. Similar distribution can be used for f(p) in (5). Here,
control over the operations of the entire network. However, σ1 and σ2 are the standard deviations for general opera-
it is required that the network entities must be aware of the tions per unit time and blockchain operations per unit time,
(o)
optimal points as well as know the slots, which may result (t) BS,i R τ −(λ +λ )t
respectively. Now, using βD,i = |S| 0
e 1 2 dt and
in redundant transmissions. In such a way, the energy of q √
the network can be conserved considerably. In general, the λk = σk ln( 2πσk fk )−2 +λk , k = 1, 2, λ = λ1 +λ2 , with

1089-7798 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2018.2883629, IEEE
Communications Letters
IEEE COMMUNICATIONS LETTERS 3

λ1 6= λ2 , and f1 6= f2 , the energy decaying in the network for which is the desired output at R0 > 0, R00 > 0 and τ > 0.
(t)
each vehicle can be observed by solving the integral (βD,i ), Optimal Solution: The solution to the energy-efficient
which gives transition considering the distributed clustering mechanism
−

√
 q √ q √
  !
2 σ2 − ln( 2πf2 σ2 )+σ1 − ln( 2πf1 σ1 ) +λ2 +λ1 τ
can be attained by performing optimal offloading and
(o) 1− e selecting new CHs whenever the energy demand of the
BS,i
. , (9) (I)
|S| √
 r √  r √  network increases. If St,i is the observation, which controls
2 σ2 − ln 2πf2 σ2 + σ1 − ln 2πf1 σ1 + λ2 + λ1
the changing of CH to optimize the transactions for
as an output. In actual practice, λ2 is unknown, however, its conserving energy, the solutions can be attained through
value can be estimated by assuming that all the operations following formulations:
in the network are synchronized, such that λ1 = λ2 and • Based on Optimal Stopping Theory (OST) [8], change
f1 = f2 . The deviation is varied to maintain a difference CH if
in regular and blockchain operations. Now, in such a case, max(St,i ) <
(I) (I)
Sexpected,i , (12)
−1 λ1 −λ1 2 λ −λ
1 2 (( ) )+( 2σ 2 )2 ) | {z }
f2 = f1 = 2πσ1 σ2 e
σ1 2 and λ2 becomes   −Pc √R
0
!!
−erf
r 
2 2σ 00
 Et R +λexpected
−2
σ2 ln (2πσ1 σ2 f1 ) − λ1σ−λ
2
1
1
+ λ1 , based on which, the
energy decaying is calculated as where stopping points are observable from Mayer, La-
2σ2
q
2 ln(2πf σ σ )−λ 2 +2λ λ −λ2 τ
−2σ1
grange and Supremum (MLS-OST) [8] for each CH at
1 1 2 1 1 1 1
(o)
BS,i σ1 − σ1 e
−
σ1 1 −2λ τ
negligible Heston variance [7], such that
.  q  . (10)
|S| 2 2 2
Z R00
!
2 σ2 −2σ1 ln (2πf1 σ1 σ2 ) − λ1 + 2λ1 λ1 − λ1 + λ1 σ1 (I) (t) 0
St,i = sup Et f (p) .Pc dp + sup f (B ) . (13)
0≤t≤τ 0 0≤t≤τ
The values for these help to form an estimated overview of
Using Lemma-1 and Lemma-2, (13) deduces to
energy requirements/consumptions of each vehicle, and allow     
0 00
tracking of differential variations for energy demands in (4).  −Pc erf R
√ −R − erf √R
0
 2σ 00 2σ 00 
R R
Such provisioning facilitates the selection of CHs with a high max Et
2
+ 2λ1 . (14)
availability of energy. In addition, the extent up to which the
• If OST does not hold, Lemma-2 is used for changing
network can survive at a given depletion rate is observable (I) (T ) (I)
CH, i.e. St,i = KR,i and if St,i <
through these results. | {z }
Lemma-2 The number of transactions needed to shift the Upper limit of handling transactions
(T )
entire load depends on the number of local as well as global KR , select CH that satisfies this condition provided that
updates performed in the IoV. The local updates can be R ≤ R00 . If no CH is available, divide load based  on
(τ )
accumulated and transferred once a slot for global exchange max (R00 ), or divide the load based on max Ft for
is initiated. Based on these, the actual number of required the given conditions in (8).
(T )
transactions (KR ) is given by • If both the above metrics fail to decide or create ambigu-
(I)
ity, St,i is modeled for pre-energy decaying, i.e. change
P 
|C| Pψ 2
erf R0 − R00 − erf R0



'
j=1 λ Pc τ
&
−
i=1
, (11) R τ −ω   (t)  √ dB0  
5
2 2 DσR00 CH, if 0 λ βD,i +  σ dt dt <
expected
R0 > 0, R00 > 0 and τ > 0 assuming that all vehicles (o)
2λ1 BS,i
follow similar pattern where R00 is their radio range, R0 |S|
.

is the average range of the network required to maintain a 

√
 q √ q √
  !
2 σ2 − ln( 2πf2 σ2 )+σ1 − ln( 2πf1 σ1 ) +λ2 +λ1 (τ −ω)
connection, σR00 is the deviation of R00 from R. 1− e
−

Proof: In the given network, the number of transactions

√ √ √
 r  r 
are driven by λ, number of parallel links D, and the 2 σ2 − ln 2πf2 σ2 + σ1 − ln 2πf1 σ1 + λ2 + λ1
connectivity function of the network G(t). Here, G(t) is !
√
driven by f (p) and Pc , such that, the number of transitions +2λ1  σ1 σ2 (τ − ω) , (15)
to be performed for the P blockchain can
 be written as
R τ R R00 |C| Pψ
1
f (p).Pc dp where ω is the previous time slot/iteration for which the
D 0 0 i=0 i=0 λ.t dt, which on fol-
lowingthe Gaussian distribution for vehicles can be written as observational values are available.
x−R0 2
  
1
R τ R R00 1 −1 2 σ
P
|C| Pψ
 The proposed optimal solution is convergent and guarantees
√ e R 00
.P dx λ.t dt.
D 0 0 2π c i=0 i=0 uniqueness at R > 0, R0 > 0, R00 > 0 and τ > 0.
Now, solving in parts, it can be written as However, the uniqueness is subject to λ1 and λ2 . As discussed
  
R 0 00
 
0
 
earlier, for practical cases, λ2 is unknown, because of which
−Pc erf √ −R − erf √R
 
|C| ψ
1  τ
the solution converges w.r.t. λ1 only. For extreme large values
2σ 00 2σ 00
Z
R R  X X
   λ.t dt.
D 2 (R → ∞, R0 → ∞, R00 → ∞), the optimal observation is

0 i=0 i=0

affected and there can be more than one solution for (9)-(14),
By 
solving the remaining
 part, the observed value becomes because of which the identification of CH becomes ambiguous
& P|C| Pψ 2 0 00 0 '
i=1 j=1 λ Pc τ (erf (R − R ) − erf (R )) and (15) is used as a solution. However, such a case is most
− 5 , unlikely to occur in practical scenarios. All the procedures
2 2 DσR00

1089-7798 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/LCOMM.2018.2883629, IEEE
Communications Letters
IEEE COMMUNICATIONS LETTERS 4

defined in the system model can be accumulated, as shown while keeping its principles intact. The performance was
in Algorithm 1, which suggests when to change the CH and recorded for the number of transactions required to offload
time slot in which pre-offloading should be performed to avoid the entire blockchain (Fig. 3) and total energy conserved in
delays and conserve the energy. The operational complexity of variation with the number of requests (Fig. 4). The results
the algorithm depends on C and ψ. The distributed and parallel in Fig. 3 show that the proposed scheme overpowers the
operations can further help to conserve the energy associated traditional blockchain by reducing the transaction load up to
with the onboard processing on each vehicle. 82.06% even at higher values for λ. These results govern the
optimal solutions for selecting CH whenever the network’s
Algorithm 1 CH Selection and offloading energy consumption goes beyond the set limits. Moreover,
(I)
1: Input: Set network metrics and set idealistic values for Sp,i the proposed scheme shows an average energy conservation
2: Set: Initial CHs and mark slots, ω, t
of 40.16% compared with traditional blockchain at varying
3: repeat
4: Compute energy ratings for each vehicle and find St,i
(I) requests. Although the conserved energy decreases with an
(I) (I)
5: Change CH if St,i < Sp,i increase in the number of requests, the increase is lower than
6: until: t ≤ τ , (7) and (8) hold
the standard operations and the conservation follows the values
(I) (I)
7: Output: Offloading at t − ω for St−ω,i < Sp,i and new CH
33.32%, 40.02%, 42.85%, 44.45% for different values of λ,
λ1 and λ2 , as shown in Fig. 4. At present, the security of the
5
x 10
network depends on the success of transactions and updates
2.5
Proposed Scheme, λ=2 shared between the vehicles. The intermittent evaluations of
Proposed Scheme, λ=3
the principles of the blockchain-security are beyond the scope
for Blockchain offloading
Number of Transactions

2 Proposed Scheme, λ=4

Proposed Scheme, λ=5

1.5
1 Traditional Blockchain, λ=5
0.5
0
10
Fig. 3. Number of Transactions vs. operational time.
IV. P ERFORMANCE E VALUATIONS
The proposed scheme was numerically evaluated for its
efficacy compared with the traditional blockchain solution,
which relies on the broadcasting of ledgers through MatlabTM .
The evaluations were carried with settings |C| = 5, |I| = 10
in each cluster, |S| = 10, λ = 2 ∼ 5, γ = λ requests
per blockchain per second, k = 3, λ1 = 1 ∼ 4, H = 10,
(t)
λ2 = 2 ∼ 5, ER = EC = 2580J, and βC,M = 0.625J is the ap-
prox. average, which can be fixed based on the curve algorithm
or any dedicated security scheme [9] [10]. Excessive energy
is taken twice as the total energy, i.e.,  = 2, τ = 10 ∼ 100s,
R = 500m [11], R0 = R00 = 300m, ω = 1s, Pc = 1
i.e. maximum presence of vehicles. In addition, standard
normal distribution was considered for normal and blockchain
operations with zero mean and unit deviation. The proposed
scheme helps to conserve energy by following a non-standard
formation of clusters against the original policies of blockchain
0.5
Excessive Energy Utilization
Traditional Blockchain, λ=2 of this article and will be discussed in future reports.
Traditional Blockchain, λ=3
Traditional Blockchain, λ=4
V. C ONCLUSION
This letter aims at reducing the number of transactions
20 30 40 50 60 70 80 90 100 required to update ledgers on the Internet of Vehicles. The pur-
Timing Slots (τ) (s)
pose of the proposed approach is to reduce the burden of the
network from numerous blockchain-transfer operations while
conserving a maximum amount of available energy. To satisfy
this goal, a distributed clustering model is designed which
helps to conserve energy by 40.16% (average) and reduces
the number of ledger-transactions by 82.06% compared with
the traditional blockchain-based Internet of Vehicles (IoV).
R EFERENCES
[1] J. Gu, B. Sun, X. Du, J. Wang, Y. Zhuang, and Z. Wang, “Consortium
blockchain-based malware detection in mobile devices,” IEEE Access,
vol. 6, pp. 12118–12128, 2018.
[2] H. Liu, Y. Zhang, and T. Yang, “Blockchain-enabled security in electric
vehicles cloud and edge computing,” IEEE Network, vol. 32, no. 3,
pp. 78–83, 2018.
[3] E. Uhlemann, “Time for autonomous vehicles to connect [connected
vehicles],” IEEE Vehicular Technology Magazine, vol. 13, no. 3, pp. 10–
13, 2018.
[4] C.-Y. Wei, A. C.-S. Huang, C.-Y. Chen, and J.-Y. Chen, “Qos-aware
hybrid scheduling for geographical zone-based resource allocation in
cellular vehicle-to-vehicle communications,” IEEE Communications Let-
ters, vol. 22, no. 3, pp. 610–613, 2018.
[5] J. Lim, H. Yu, K. Kim, M. Kim, and S.-B. Lee, “Preserving location
privacy of connected vehicles with highly accurate location updates,”
IEEE Communications Letters, vol. 21, no. 3, pp. 540–543, 2017.
[6] X. Liu, W. Wang, D. Niyato, N. Zhao, and P. Wang, “Evolutionary
game for mining pool selection in blockchain networks,” IEEE Wireless
Communications Letters, pp. 1–1, 2018.
Proposed Scheme [7] E. Benhamou, E. Gobet, and M. Miri, “Time dependent heston model,”
Traditional Blockchain
SIAM Journal on Financial Mathematics, vol. 1, no. 1, pp. 289–325,

Average Energy Conservation

2010.
Factor (Lemma−1)

0.4 by the proposed approach= 40.16%

[8] G. Peskir and A. Shiryaev, Stochastic processes: A brief review, pp. 53–
0.3
121. Basel: Birkhäuser Basel, 2006.
0.2 [9] A. Liu and P. Ning, “Tinyecc: A configurable library for elliptic curve
cryptography in wireless sensor networks,” in International conference
0.1
on Information processing in sensor networks, pp. 245–256, IEEE, 2008.
0
[10] N. R. Potlapally, S. Ravi, A. Raghunathan, and N. K. Jha, “Analyzing the
C1 C2 C3 C4
Sceanarios energy consumption of security protocols,” in International symposium
(C1: λ1=1, λ2=λ=2; C2: λ1=2, λ2=λ=3; C3: λ1=3, λ2=λ=4; C4: λ1=4, λ2=λ=5)
on Low power electronics and design, pp. 30–35, ACM, 2003.
[11] U. Ozguner, T. Acarman, and K. A. Redmill, Autonomous ground
Fig. 4. Energy conservation factor (Lemma-1) vs. requests per unit time. vehicles. Artech House, 2011.

1089-7798 (c) 2018 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.