IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO.

1, FEBRUARY 2023 3

Multi-Operator Dynamic Spectrum Sharing for

Wireless Communications: A Consortium
Blockchain Enabled Framework
Zuguang Li, Wei Wang , Member, IEEE, Qihui Wu , Senior Member, IEEE, and Xianbin Wang , Fellow, IEEE

Abstract—To enable secure and efficient dynamic spectrum frequencies to millimeter wave (mmWave), terahertz (THz),
sharing (DSS) with guaranteed revenue and quality of service and optical bands [2].
(QoS) in future wireless communications, we present a consor- While 6G is evolving towards higher frequencies, efficient
tium blockchain based DSS framework, where the regulators
supervise the whole process of DSS, and thus the revenue of each spectrum management at the sub-6 GHz band still remains
participant can be guaranteed. Each mobile network operator essential due to its superior indoor propagation property.
(MNO) on the chain can adaptively act as a spectrum provider Disregard the golden spectrum resources at the sub-6 GHz
or spectrum requestor based on their demand, and the spectrum band are almost exhaustively licensed, it has been reported that
resource allocation is recorded on the chain with a smart contract. the actual spectrum utilization rate of this band is only around
The optimal spectrum pricing and buying strategies are solved
based on a multi-leader multi-follower (MLMF) Stackelberg 15% to 85% [3], which is far from an ideal scenario. The
game model, and the equilibrium is solved with the proposed spectrum shortage and its low utilization rate have motivated
algorithm. We then build a prototype with Hyperledger Fabric researchers to develop dynamic spectrum sharing (DSS) tech-
consortium blockchain, and the average latency is evaluated. nologies, which have been very active research topics over the
Simulations and prototype evaluations validate the feasibility of past two decades [4], [5], [6], [7], [8]. Despite the development
blockchain based DSS and show that the average latency increase
with the participants, which provides useful insights for real of numerous DSS solutions since the emergence of cognitive
applications. radio (CR) technologies [9], like the licensed shared access
(LSA) system [10] and the spectrum access system (SAS) for
Index Terms—Multi-operator spectrum sharing, consortium
blockchain, smart contract, Stackelberg game. citizens broadband radio service (CBRS) system [11], large-
scale DSS is still not deployed in practice. One main obstacle
lies in the lack of reasonable incentives for spectrum resource
I. I NTRODUCTION providers, guaranteed quality of service (QoS) for secondary
users, and security and privacy protection for both partici-
ITH the ongoing deployment of the fifth generation
W (5G) wireless infrastructure, the research for the next
sixth-generation (6G) wireless networks has started, with a
pants, which directly results in the reluctance for sharing of
previously licensed spectrum resources.
As an emerging technology with unique property of trans-
focus to support diverse smart applications, such as virtual and
parency, anonymity, immutability, and traceability, blockchain
augmented reality, holographic communications, ultra-high-
has been regarded as a promising tool to empower DSS
resolution video streaming, etc [1]. With such an ambitious
in future wireless communication networks [12], [13], [14].
goal, 6G is anticipated to adopt both evolutionary and dis-
Leveraging the cryptographic and signature techniques, the
ruptive technologies, including space-air-ground integrated
identity of a participant in a blockchain network can be
networking for global ubiquitous coverage, and utilization of
verified by others, but it does not need to provide any
full spectrum ranging from low and mid-band at sub-6 GHz
private information. With the multi-party jointly maintained
Manuscript received 12 May 2022; revised 25 August 2022; accepted distributed ledgers, the spectrum assets ownership, right to
27 September 2022. Date of publication 6 October 2022; date of current ver- use and corresponding sharing process can be recorded per-
sion 8 February 2023. This work was supported in part by the National Key manently on the chain, and thus the revenue of both the
R&D Program of China under Grant 2020YFB1005900, the National Natural
Science Foundation of China under Grant 62001220, the Natural Science spectrum resource provider and the resource consumer can be
Foundation of Jiangsu Province BK20200440, the Future Network Scientific guaranteed. Moreover, any improper spectrum usage, such as
Research Fund Project FNSRFP-2021-YB-03, and the Young Elite Scientist breaking the power and bandwidth limitations, can be iden-
Sponsorship Program, China Association for Science and Technology. The
associate editor coordinating the review of this article and approving it for tified with the assistance of off-chain spectrum monitoring.
publication was Y. Pei. (Corresponding author: Wei Wang.) Particularly, with the deployed smart contract and spectrum
Zuguang Li, Wei Wang, and Qihui Wu are with the College of assets information on the chain, spectrum allocation decision
Electronic and Information Engineering, Nanjing University of Aeronautics
and Astronautics, Nanjing 211106, China (e-mail: [email protected]; can be made automatically and reach a consensus with mutual
[email protected]; [email protected]). interference avoidance [15]. Compared with the conventional
Xianbin Wang is with the Department of Electrical and Computer centralized database based approaches, the blockchain based
Engineering, Western University, London, ON N6A 3K7, Canada (e-mail:
[email protected]). approach shows great potential to facilitate secure and efficient
Digital Object Identifier 10.1109/TCCN.2022.3212369 DSS, and it has been considered as a promising solution by
2332-7731 
c 2022 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See https://www.ieee.org/publications/rights/index.html for more information.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
4 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

the Federal Communications Commission (FCC) of the United frequency band. DSS was also included in Release 16 [31]
States at the 2018 Mobile World Congress (MWC) [16]. and further enhanced in Release 17.
It is worth noting that even though existing works [14], There have been some works considering to use blockchain
[17], [18], [19], [20], [21], [22], [23], [24], [25], [26], [27], for DSS in the literature [14], [17], [18], [19], [20], [21], [22],
[28], [29] have been discussed to use public or consortium [23], [24], [25]. Blockchain as a platform for spectrum allo-
blockchain for DSS, the detailed operations are still not cation was proposed in [18], where secondary users (SUs)
clear, and most of them directly follow a framework sim- obtain the spectrum access right through buying the idle spec-
ilar to CR, which cannot be applied in practice. Different trum of primary users (PUs). Combining ring signature, a
from [26] and [28], in this paper, we consider operator level privacy protection mechanism was proposed to protect nodes’
spectrum sharing, where the regulators supervise the whole identities when they upload data. Likewise, a blockchain-
process of DSS. Each micro or virtual mobile network operator based proof-of-strategy (PoS) consensus scheme for DSS
(MNO) can adaptively act as spectrum provider or spectrum between PUs and SUs was proposed in [19], where the ring
requestor based on their demand, and we design a multi-leader signature technique was designed as the privacy protection
multi-follower (MLMF) Stackelberg game to achieve optimal method for users. In [20], a blockchain-enabled reverse auc-
spectrum pricing and resource allocation to maximize the rev- tion mechanism was devised for radio access network (RAN)
enue of all participants. The Stackelberg equilibrium of the sharing and dynamic users’ service provision in beyond 5G
game is analysed and is solved through the best response (B5G) networks, where a public blockchain is used to record
approach. Moreover, we build a complete simulation envi- spontaneous service requests made by users, while a pri-
ronment with Hyperledger Fabric to evaluate the performance vate blockchain allows operators and other participants to
of DSS. The contributions of this paper are summarized as share spectrum resources. However, most of them directly fol-
follows: low a framework similar to CR, which cannot be applied
• We propose a consortium blockchain-based DSS frame- in practice. In [17], the authors constructed a blockchain
work, where the regulators are responsible for reg- based decentralized self-organized trading platform for indus-
ulating the whole process of DSS, which facilitates trial Internet of Things (IIoT) devices, where the interactions
the security and efficiency of DSS among different between the cloud provider and miners are modeled as a
operators. Stackelberg game, which assists the lightweight IoT devices
• We model the interactions between multiple buyers and to offload huge computational tasks to the cloud provider.
sellers as a MLMF Stackelberg game, where micro or In [14], the authors proposed a two-tier hierarchical blockchain
virtual MNOs can dynamically share or rent spectrum empowered dynamic resource sharing architecture, where a
resources for their own demand. The optimal spectrum public blockchain is maintained and accessed by all the users,
pricing and resource allocation is optimized to maximize while a private blockchain supports the DSS for base stations.
the revenue of all participants, with the constraints of Similarly, a blockchain-based spectrum access system com-
available bandwidth of each seller operator (SOP), and prising of two layers of blockchain networks was proposed
budget of each buyer operator (BOP). The Nash equi- in [21], where a global chain is used to serve the servers and
librium is proved to exist and the optimal solution is regulator nodes at a global scale, and a local chain is dedicated
obtained with the proposed algorithm. for spectrum access management for a local spectrum zone.
• We design a smart contract incorporating the MLMF In public blockchain (e.g., Bitcoin [32] and Ethereum [33]),
Stackelberg game, and build a prototype with anyone is free and anonymous to join and participate in the
Hyperledger Fabric consortium blockchain to ver- core activities of the blockchain, which is unfavorable for
ify the feasibility of our proposed DSS system. The the regulators to supervise DSS. As a counterpart of pub-
average latency performance of the blockchain is lic blockchain, consortium blockchain can only accept the
evaluated. approved participants, and thus can be applied to facilitate
The remainder of this paper is organized as follows. We DSS with regulation. In [26], the authors devised a con-
discuss related works in Section II and present the system sortium blockchain framework to achieve spectrum sharing
model in Section III. The MLMF Stackelberg game is formu- in multi-operator wireless communication networks, where a
lated in Section IV, followed by the equilibrium analysis and double auction and free-trading market based on smart con-
solutions in Section V. In Section VI, simulation results are tract was designed to enable operators to autonomously share
presented, and the performance analysis are also discussed. the spectrum resource. However, the spectrum auction algo-
Finally, conclusions are given in Section VII. rithm is finished by matching a bidding queue or comparing
whether the lowest seller price is higher than the highest buyer
price, which is not an optimal strategy and does not provide
II. R ELATED W ORKS an effective incentive for operators to share their spectrum.
Recently, DSS among multiple operators has been intro- In [27], a spectrum-efficient framework based on consortium
duced by the industry and academia to enable wireless com- blockchain was proposed for human-to-human and machine-
munication services. In 3GPP Release 15 [30], DSS as a to-machine coexistence in 5G heterogeneous networks, where
technology was proposed to provide a migration path from PUs (i.e., human-to-human users) can share their unused spec-
fourth generation (4G) long term evolution (LTE) to 5G new trum resources with SUs (i.e., machine-to-machine users).
radio (NR) by allowing LTE and NR to share the same In [28], to provide high speed network services with reliable

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 5

Fig. 1. An illustration of consortium blockchain enabled DSS system.

QoS, a blockchain-based framework was proposed to evaluate Then, the BSs send the spectrum demand request to the cor-
5G spectrum sharing and model intra and inter spectrum man- responding operation administration and maintenance (OAM)
agement among multiple telecom-operators. However, how to server of each operator, which is responsible for spectrum
achieve spectrum sharing is not clearly presented. In [29], a assignment and management [26]. According to the request
spectrum trading blockchain model was presented to improve information collected from the BSs, each OAM server can
the efficiency and security of spectrum transactions in vir- determine the spectrum demand during the slot.1 For the ease
tual optical networks, where a new consensus solution was of understanding, we summarize the notations in Table I.
proposed to quickly confirm transactions, and a temporary To facilitate secure and efficient spectrum sharing, the
anonymous transaction was designed to prevent mainstream operators form a consortium blockchain, where each opera-
attacks. In [34], a hierarchical blockchain and deep reinforce- tor stands for an organization. The operators can choose to
ment learning framework was proposed to achieve secure become seller operators (SOPs) or buyer operators (BOPs)
resource trading in 5G RAN, where a single-leader multi- based on their spectrum demand results at the beginning of
follower Stackelberg game was formulated to determine the each slot, and they are divided into SOPs group and BOPs
buyers’ demands and the seller’s price, and a consortium group. The SOPs group consisting of N operators is denoted
blockchain network was deployed to support their resource as N = {n|n = 1, . . . , N }. The total available spectrum
trading smart contract. resources of SOP n for selling is Bn , with a unit price pn . The
pricing strategies of all SOPs are denoted as p = [p1 , . . . , pN ].
III. DSS S YSTEM W ITH C ONSORTIUM B LOCKCHAIN Similarly, the BOPs group consisting of M operators is
In this section, we first introduce the DSS system, where denoted as M = {m|m = 1, . . . , M }. We assume that
multiple operators form a consortium blockchain to achieve the total budget of BOP m is Qm with a buying strategy
secure and efficient inter-operator spectrum resource sharing. bm = [bm,1 , . . . , bm,N ], where bm,n is the bandwidth bought
Then the operation details of DSS are described. by BOP m from SOP n. The buying strategies of all BOPs are
denoted as B = [b1 , . . . , bM ].
For the consortium blockchain, each organization consists
A. System Overview
of a client, some peer nodes, and orderer nodes. Peers are
As shown in Fig. 1, we consider a DSS scenario in a specific responsible for executing spectrum sharing strategy and trans-
region, where each operator has one macro base station (MBS) actions while orderers package the transactions and generate
and several small base stations (SBSs). Different types of user a new block for consensus [21]. To ensure the secure and
terminals (e.g., vehicles, unmanned aerial vehicles (UAVs), robust operation of DSS, there are regulators in the consortium
and smart phones, etc.) are connected to the corresponding
nearby BSs for wireless services. We assume that the DSS is 1 The OAM server can calculate the terminals’ spectrum demand by two
conducted among different operators in a slot-by-slot manner. methods: (1) The terminals report their spectrum demand to the associated
The BSs monitor spectrum usage within the region, and ana- MBS or SBSs, then the OAM server counts the total spectrum demand col-
lected from the MBS and SBSs; (2) By leveraging the prediction tools, the
lyze spectrum demand at the beginning of each slot. Without OAM server predicts the terminals’ spectrum demand based on the historical
loss of generality, we assume that the duration of each slot is T. spectrum usage in the specific region.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
6 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

TABLE I
N OTATION

Fig. 2. Flowchart of DSS between BOPs and SOPs.

B. Operation Details of DSS

In the DSS system, elliptic curve encryption algorithm is
blockchain, which are responsible for regulating the process used to generate digital significance for each organization. In
of DSS and managing the permissions of nodes. The main order to obtain effective supervision, each operator must first
duties of regulators are listed as follows: be certified by a certificate authority (CA) before it become a
1) The regulators synchronize the data on the chain, and legitimate organization, and it is licensed to a digital identify
then regulate and audit the whole process of DSS (ID) based on its organizational role. Legal organization will
through mining and analyzing these data. To prevent now receive public-private key pairs and corresponding cer-
the occurrence of risk, the regulators trace the history of tificates for encrypting shared data. When sharing spectrum
each transaction and monitor the transaction information with other operators, the operator specifies and presents its
of the associated nodes. If the regulators identify vio- ID. A membership service provider governs the valid ID for
lations, they will have the right to revise and delete the operator.2 As shown in Fig. 2, the DSS flow consists of
corresponding transactions in that blocks. nine phases:
2) The regulators are responsible for the certificate autho- 1) On the BOP side, it signs and transmits its spectrum
rization of operators before they join the blockchain. request message MSGreq to the memory pool via client, which
To monitor the running status of nodes, the regulators is expressed as follows:
can dynamically modify the permissions of nodes, such MSGreq = {idm , sigm , locm , Qm , feem , slotno }, (1)
as consensus and transaction. If some nodes are found
malicious, the regulators will remove them. where idm , sigm , and locm are ID, digital significance, and
Moreover, smart contract is immutable once deployed, location information of BOP m. feem is the prepaid coin
which guarantees immutability. To achieve automatic exe- of BOP m for spectrum sharing, which contains the trans-
cution of the spectrum sharing strategy between BOPs and action fees for the primary node processing the transactions
SOPs, the DSS smart contract is deployed on the consor- and deposit for the violation. slotno is the slot number.
tium blockchain. The execution of the DSS smart contract is Similarly, to share the available spectrum resource, SOP n
restricted to primary nodes that vouch for the results of spec- signs and sends its selling message MSGsel to the memory
trum sharing strategy. To reduce the communication overhead, pool, which is denoted as:
one primary node is selected to execute the DSS smart con- MSGsel = {idn , sign , locn , Bn , Cn , feen , slotno }, (2)
tract based on reputation. The higher reputation of the peer
node, the more likely it is to be selected as the primary node. where Cn is SOP n’s cost coefficient per unit bandwidth.
Furthermore, the primary node will get reward for executing 2 https://hyperledger-fabric.readthedocs.io/en/release-
the DSS smart contract after finishing spectrum sharing. 2.2/identity/identity.html.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 7

2) The selected primary node collects these messages from directly related to the provided bandwidth and is given by
the memory pool. Cn bm,n , where Cn is the cost per unit bandwidth. Therefore,
3) The primary node automatically executes the DSS smart for SOP n, the revenue function can be defined as follows:
contract to compute a spectrum sharing strategy.

M
4) The primary node packages the spectrum messages and Rn = (pn − Cn )bm,n . (3)
the spectrum sharing strategy into a transaction proposal (i.e., m=1
proposal) signed with its signature, and then as a client uploads
For BOPs, the more bandwidth they buy, the larger utility
the proposal to endorsing peers (endorsers).
they may receive. The utility should be an increasing function
5) The endorsers simulate the proposal by executing the
of the bought bandwidth. Hence, for BOP m, a log function
operation on another specified smart contract. After the simu-
can be used to model the utility which can be defined as:
lation, the endorsers cryptographically sign a message called
endorsement, which contains readset and writeset (together 
N
  N
with metadata such as transaction ID, endorser ID, and Um = w m log 1 + ηm bm,n − bm,n pn , (4)
endorser signature), and send it back to the primary node n=1 n=1
in a proposal response. The primary node collects endorse- where wm denotes the weight of BOP m in current DSS, which
ments until they satisfy the endorsement policy, which means is a positive real number. ηm is the transmission channel gain
that most endorsers agree on the transaction, and then pro- of BOP m, and ηm ∈ [0, 1].
ceeds to create the transaction and passes it to the ordering
service. Ordering nodes (orderers) establish the total order of B. Problem Formulation
all transactions, batch multiple transactions into blocks, and
Given the utility functions of SOPs and BOPs, we can for-
output a hash-chained sequence of blocks containing transac-
mulate the optimal spectrum buying and pricing problems for
tions. But they do not participate in the execution nor in the
the lower and upper layers of the MLMF Stackelberg game.
validation of transactions. Then the orderers send blocks back
The lower-layer game can be defined as follows.
to the endorsers. Finally, the endorsers validate blocks includ-
Problem 1: The problem in the lower layer (followers’ side)
ing the identities of primary node, parameters, the identifier
can be formulated as:
of the DSS smart contract, and signatures on endorsements of  
the transaction. max Um bm,n , B−m , p
6) Peers update the spectrum sharing ledger using the bm,n
blocks [35]. s.t. bm,n ≥ 0, ∀m ∈ M, ∀n ∈ N ,
7) BOPs and SOPs obtain the spectrum sharing strategy 
N
from their ledgers. pn bm,n ≤ Qm , ∀m ∈ M,
8) After that, the rewards for the primary node are paid by n=1
the transaction fees from both BOPs and SOPs. M
9) BOPs pay SOPs for buying bandwidth after receiving bm,n ≤ Bn , ∀n ∈ N . (5)
the spectrum sharing strategy. If BOPs or SOPs violate the m=1
spectrum sharing rule, their deposits will be confiscated. Since the other BOPs’ buying strategies B−m affect the
SOPs’ pricing strategies p, the utility function Um is an
IV. MLMF S TACKELBERG G AME implicit function of B−m . The first constraint in (5) means
that the bandwidth purchased from SOPs is nonnegative. The
In this section, we first give the utility functions of BOPs
second constraint guarantees that the cost of BOP m’s buying
and SOPs, and then describe their problems. Specifically, we
bandwidth cannot exceed its budget Qm . The third constraint
model the interactions between BOPs and SOPs as a MLMF
guarantees that the bandwidth which SOP n sells to all BOPs
Stackelberg game.
cannot exceed its own available bandwidth Bn .
At the upper layer, the optimization problem of SOP n can
A. Utility Function be defined as follows.
Each participant aims to maximize its revenue driven by Problem 2: The problem in the upper layer (leaders’ side)
its selfishness. SOPs decide the pricing strategies to maximize can be formulated as:
their revenues based on the budgets of BOPs and their own
bandwidth constraints. BOPs determine the buying strate- max Rn (pn , p−n , B)
p n
gies given the pricing strategies of the SOPs. Therefore, the s.t. pn ≥ Cn , ∀n ∈ N ,
interactions between SOPs and BOPs can be modeled as a
pn ≤ P n , ∀n ∈ N . (6)
MLMF Stackelberg game, where SOPs and BOPs act as lead-
ers and followers, respectively. We consider the reasonable The first constraint in (6) guarantees that the utilities
utility functions of SOPs and BOPs, so that the pricing and of SOPs are nonnegative. The second constraint means
buying strategies can be obtained. that the price pn cannot exceed its maximum value Pn .
For SOPs, the revenue function can be defined as the income Considering the scarcity of spectrum, the less bandwidth
by providing bandwidth to BOPs minuses its service cost and SOPs have, the higher prices they may set. The maximum
charge for the primary node. The service cost of SOP n is price Pn should be related with Bn . Suppose that Pn =

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
8 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

φn exp(−Bn / max{B1 , . . . , BN }), where φn is the weight where wm ≥ 0 is the profit coefficient of BOP m.
factor. Consequently, we have ∂ 2 Um /∂bm,n 2 ≤ 0. Thus, the utility
For a MLMF Stackelberg game, its Stackelberg equilibrium function Um is a concave function with respect to bm,n , and
is the Nash equilibrium between the leaders and followers. The the problem for each BOP in the lower-layer game is a convex
Nash equilibrium of a game is a point at which no player optimization problem. The interaction among the BOPs forms
(leader or follower) can unilaterally increase its utility by a concave M-person game. Therefore, the Nash equilibrium
changing its strategy without deteriorating the utilities of the point of the lower-layer game always exists and is unique.
other players [36]. In this case, the Nash equilibrium point is The first order derivative of the utility function Um should
expressed by the best response that is the optimal strategy of satisfy (∂Um /∂bm,n )|bm,n =0 ≥ 0, ∀m ∈ M. Otherwise,
one player, which is defined as follows: BOPs have no motivation to buy bandwidth from SOPs
Definition 1: Let B∗ and p∗ be the optimal strategies because of no profit. It means that pn ≤ P̄ , where P̄ =
of BOP and SOP, respectively. The point (B∗ , p∗ ) is the min{w1 η1 , . . . , wM ηM }. We have Cn ≤ pn ≤ Pn based on
Stackelberg equilibrium of the MLMF Stackelberg game if the constraints in (6). Therefore, the pricing strategy space of
it satisfies the following conditions. For each BOP m, SOP n is [Cn , min{Pn , P̄ }], which is bounded. According
 ∗    to Proposition 1, the lower-layer game can be decom-
Um bm,n , B∗−m , p∗ ≥ Um bm,n , B∗−m , p∗ , ∀m ∈ M, (7)
posed to M distributed optimization problems. Sequentially,
and for each SOP n, we use the Karush-Kuhn-Tucker (KKT) conditions to solve
   
Rn pn∗ , p∗−n , B∗ ≥ Rn pn , p∗−n , B∗ , ∀n ∈ N , (8) Problem 1, and its optimal solution is obtained in the following
proposition.
where bm,n and pn are arbitrary feasible strategies for the Proposition 2: For a given pricing strategy p, the optimal
BOP and the SOP, respectively. solution for Problem 1 is given by:
⎧
⎪ 0, if pn ∈ C1,
V. S OLUTION OF THE MLMF S TACKELBERG G AME ⎪ l
⎨
∗ bm,n , if pn ∈ C2,
In this section, we first analyze the existence and uniqueness bm,n = (13)
⎪
⎪ b r , if pn ∈ C3,
of the Nash equilibrium in terms of the lower-layer and upper- ⎩ wm,n 1
pn − ηm , else,
m
layer problems, respectively. Then, we analyze the solution of
the lower-layer problem, and leverage the root mean square where C1, C2, and C3 are defined in (25), (27), and (29),
propagation (RMSProp) algorithm to solve the upper-layer respectively.
problem. Proof: Please refer to Appendix A.

A. Lower Layer (Followers’ Side) Analysis

B. Upper Layer (Leaders’ Side) Analysis
For the lower-layer game, each BOP decides its buying strat-
egy based on the SOPs’ pricing strategies p. We have the On the leaders’ side, each SOP determines its unit price
following proposition about the lower-layer Nash equilibrium. pn to maximize its revenue Rn . Note that the pricing strate-
Proposition 1: The Nash equilibrium point of the lower- gies of SOPs directly affect the buying strategies of BOPs.
layer game exists and is unique, regardless of SOPs’ pricing Consequently, the pricing strategies of SOPs affect each
strategies p. other’s revenue. The upper-layer game among SOPs is non-
Proof: As defined in Problem 1, the strategy space of BOP cooperative and competitive. In the following, we will analyze
max ] which is nonempty, closed, and convex, and
m is [0, bm,n the Nash equilibrium of the upper-layer game, and design an
max
bm,n is defined as: algorithm to find its solution.
  For the upper-layer game, we first prove the existence and
max l r
bm,n = min bm,n , bm,n , (9) uniqueness of its Nash equilibrium point.
Proposition 3: The Nash equilibrium point of the upper-
where bm,n
l and bm,n
r are defined as: layer game exists and is unique.
⎧  Proof: Please refer to Appendix B.
⎨ l Qm − N p b
n  =n n  m,n 
bm,n = pn , (10)
Given the pricing strategies of all leaders, the followers will
⎩ b r = B − M determine their buying strategies. For the MLMF Stackelberg
m,n n m =m bm  ,n .
game, we have the following proposition.
Therefore, the lower-layer game exists a Nash Proposition 4: The Stackelberg equilibrium point of the
equilibrium [37]. MLMF Stackelberg game exists and is unique.
The first and second order derivatives of the utility function Proof: As analyzed in Proposition 2, each BOP can find
Um with respect to bm,n are calculated as: its optimal buying strategy, which is unique given the pricing
∂Um wm ηm strategies. Each SOP in the upper-layer game has a unique
= − pn , (11)
∂bm,n 1 + ηm bm,n Nash equilibrium point, based on Proposition 3. Thus, we can
conclude that the Stackelberg equilibrium point of the MLMF
and
Stackelberg game exists and is unique.
∂ 2 Um wm ηm2
The Nash equilibrium point of the lower-layer game can
2
= −  2 , (12)
∂bm,n 1 + ηm bm,n be found based on Proposition 2, given in (13). However,

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 9

the Nash equilibrium point of the upper-layer game cannot Algorithm 1 Optimal Pricing and Demand Optimization
be expressed in a closed form, and can only be calculated Algorithm
through numerical manners. Next, we leverage the RMSProp Input: The number of iterations I, budgets of all BOPs {Qm },
algorithm with an adaptive learning rate to find the numeri- initial prices of all SOPs {pn }, adaptive learning rates
cal solution of the upper-layer game, which is an extension {δi }, constants ε and ξ, and momentum parameters {γi };
of gradient descent in order to locate the optimization of an Output: The buying strategies B for BOPs, and the pricing
function. In the RMSProp algorithm, the gradient of the objec- strategies p for SOPs;
tive function gradually approaches or becomes zero based on 1: // Initialize U, R, J, K, j = 0, and k = 0;
the cyclic iteration scheme. At the i-th iteration, the pricing 2: for i = 1 to I do
strategy of SOP n is defined as: 3: // Calculate the buying strategies at the lower-layer
4: game;
δi gn,i
pn,i+1 = pn,i − 2 , (14) 5: while ||U − Uold || > ξ and j < J do
gn,i + ε 6: Uold = U;
where δi is the adaptive learning rate at the i-th iteration. 7: for m = 1 to M do
ε (ε > 0) is a constant which avoids division by zero. gn,i 8: for n = 1 to N do
denotes the partial derivative of the objective function of SOP 9: Calculate bm,n according to (13);
n at the i-th iteration, defined as follows: 10: B[m][n] = bm,n ; // B updates bm,n ;
  11: end for
∂Rn pn,i , B, p−n,i 12: Calculate Um according to (4);
gn,i = . (15)
∂pn,i 13: U[m] = Um ; // U updates Um ;
14: end for
Then, the optimal solution of the upper-layer game can be
15: j = j + 1;
solved through several iterations. The detail of the optimal
16: end while
pricing and demand optimization algorithm is shown in
17: // Calculate the pricing strategies at the upper-layer
Algorithm 1. We have the following lemma for the RMSProp
18: game;
algorithm.
19: while ||R − Rold || > ξ and k < K do
Lemma 1: The √ RMSProp algorithm will converge at a rate 20: Rold = R;
of O (log(Θ)/ Θ), where Θ is the maximum rounds of
21: for n = 1 to N do
iteration.
22: Calculate gn according to (15);
Proof: The convergence of the RMSProp algorithm has been
23: Calculate pn according to (14);
proved in [38]. The upper-layer game among SOPs is conver-
24: p[n] = pn ; // p updates pn ;
gent by using a batch size as large as the number of√maximum
25: Calculate Rn according to (3);
rounds Θ, and its convergence rate is O (log(Θ)/ Θ) under
26: R[n] = Rn ; // R updates Rn ;
satisfying the necessary restrictions listed in [38].
27: end for
28: k = k + 1;
VI. S IMULATION R ESULTS 29: end while
In this section, we focus on the spectrum sharing in the con- 30: if ||U − Uold || < ξ and ||R − Rold || < ξ then
sortium blockchain to verify the feasibility of our design. We 31: break
first show the numerical results to validate the theoretical anal- 32: end if
ysis and evaluate the performance of the MLMF Stackelberg 33: end for
game. Then, we evaluate the performance of spectrum sharing 34: return (B, p)
in Hyperledger Fabric.3

A. Evaluation of the MLMF Stackelberg Game

are assumed to 1, i.e., Cn = 1, ∀n ∈ N , similar to the config-
We consider M BOPs and N SOPs in the spectrum sharing
uration in [8]. The weight factor of SOP n is set to φn = 10.
market, as illustrated in Fig. 1. We assume that SOPs provide
The BOPs’ budgets are set to follow a Poisson distribution
the same spectrum leasing service for all BOPs, which means
with a mean value of 30, i.e., Qm ∼ Poisson(30). The BOPs’
that the buying strategies of BOPs are determined only by
profit coefficients {wm } are set to follow an uniform distribu-
their prices. Parameters in the RMSProp algorithm are set as
tion between 10 and 20. Similarly, the bandwidth efficiencies
δi = 0.001, ε = 10−6 , and γi = 0.9 which is similar to [39],
{ηm } are set to follow an uniform distribution between 0.7
at each iteration.
and 0.9.
First, we analyze the convergence of the MLMF Stackelberg
As shown in Fig. 3, the price of each SOP gradually
game, where the bandwidths provided by 5 SOPs are {Bn } =
increases at first, and then they reach their plateaus after about
{20, 40, 60, 80, 100} MHz. The initial prices of all SOPs are
10 iterations, which means that the upper-layer game reaches
set as 2, i.e., pn = 2, ∀n ∈ N . All cost coefficients of SOPs
the Nash equilibrium. We can observe that the less bandwidth
3 We have uploaded all the simulation codes and data to GitHub at SOP owns, the higher price it will set, which is straightfor-
https://github.com/lizuguang/DSS. ward and follows the spectrum trading market. In Fig. 4, the

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
10 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

Fig. 3. The pricing strategy of each SOP. Fig. 6. Revenue of each SOP in the proposed algorithm versus in a fixed
pricing approach, where SOP 2 and SOP 4 provide 40MHz and 80MHz,
respectively.

respectively. The unit prices and cost coefficients of SOPs

remain the same as above. The budgets of BOPs follow
Qm ∼ Poisson(20). For BOPs, the profit coefficient is 10,
i.e., wm = 10, and the bandwidth efficiency is 0.8, i.e.,
ηm = 0.8, ∀m ∈ M. We select one BOP and vary its budget
from 0 to 40, with a step size 2. As illustrated in Fig. 5, the
BOP can not buy bandwidth from SOP 1 with the highest price
when the BOP’s budget is very low. The BOP’s bandwidth
bought from each SOP increases gradually when its budget is
less than 10. After that, the BOP’s bought bandwidth keeps
steady, which is because of the limited bandwidth of SOPs.
Then, we evaluate the revenue of each SOP in the proposed
Fig. 4. The revenue of each SOP. algorithm and compare with a fixed pricing approach. In this
simulation, the parameters except the budget Qm are the same
as in the first simulation. All BOPs’ budgets are identical and
we vary each BOP’s budget Qm from 0 to 40, with a set size
4. The price of each SOP is the initial price in the fixed pricing
approach. As illustrated in Fig. 6, when the total budgets of
BOPs is less than 400, the revenue of SOP 2 in the proposed
algorithm is lower than that in the fixed pricing approach. It
suggests that a SOP with a high price obtains a lower revenue
in the proposed algorithm, when there are abundant spectrum
resources. However, the revenue of each SOP in the proposed
algorithm is more than that in the fixed pricing approach, when
the BOPs’ demand of spectrum resource exceeds SOPs’ sup-
ply, e.g., total budgets of BOPs more than 600. Besides, each
SOP reaches its maximum revenue faster in the fixed pricing
approach. Because it is easier to run out of bandwidth under
Fig. 5. Bought bandwidth of one BOP versus its budget, where B1 = 40, a fixed price, when the bandwidth demand is high.
B2 = 80, and B3 = 120, respectively. Finally, we compare the convergence time of the proposed
algorithm with Greedy-SM and Random-SM mechanisms
proposed in [40], which are based on the greedy algorithm to
revenue of each SOP increases and then fluctuates at first, match buyers with sellers. The convergence time is the time
because the update of a SOP’s pricing strategy will affect other from the beginning of an algorithm to the completion of its
SOPs’ utilities. In addition, we can obtain that the SOP own- convergence. As shown in Fig. 7, we can observe that the
ing more bandwidth sets a lower price, but it still achieve a convergence time of the proposed algorithm is shorter than
higher revenue. Greedy-SM and Random-SM. This is because the proposed
After that, we consider a 3 SOPs and 30 BOPs cases, where algorithm always maximizes the utilities of BOPs and SOPs
SOPs provide 40MHz, 80MHz, and 120MHz bandwidth, with simple judgements and calculations according to (13)

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 11

Fig. 8. Data instance of block header created in the DSS system, where the
data hash of the transaction information is listed in the red box.

Fig. 7. Convergence time versus the number of BOPs, where the number of select the Fabric’s native Kafka consensus, an efficient con-
BOPs is tenfold as many as SOPs. sensus protocol, for the ordering service. In the blockchain
network, the batch timeout is fixed to 2 seconds (s), the batch
TABLE II size is set to 32KB, and the maximum message count is fixed
C ONFIGURATIONS OF L INUX D ESKTOP
to 10 transactions.4 According to the regulations in Fabric, a
block is synchronized to distributed ledgers of organizations
in three cases after the consensus peers reach a consensus on
the block: (1) The delay of block synchronization triggers the
maximum synchronization timeout, which is set to 1s in our
designed blockchain network; (2) The number of blocks stored
in a cache triggers the maximum block number, which is set
to 30 in our designed blockchain network; (3) The data size
in the cache triggers the maximum data size, which is set to
512KB in our designed blockchain network.
Our evaluations focus on the core task of spectrum shar-
ing (i.e., running the DSS smart contract and the primary
node sending out the buying and pricing strategies to con-
and (14), while Greedy-SM and Random-SM obtain the local sensus peers), as it most directly impacts the spectrum sharing
optimal utilities of operators with complex calculations. system’s QoS. Specifically, two metrics are evaluated: submis-
From the above simulations, we can observe that the game sion time, which is the time between when the primary node
reaches the Nash equilibrium after 10 iterations, but the num- starts to calculate the buying and pricing strategies after col-
ber of iterations I can be set to 20 in the DSS smart contract lecting spectrum sharing requests (i.e., step 3 in Fig. 2) and
for ensuring the convergence of the game. The buying and when each endorsing peer in each organization receives the
pricing strategies formulated in the MLMF Stackelberg game strategies (i.e., step 4 in Fig. 2), and latency, which is the time
are rational to BOPs and SOPs in both abundant and short between the buying and pricing strategies calculated by the pri-
spectrum scenarios. mary node (i.e., step 3 in Fig. 2) and all operators receiving
the buying and pricing strategies (i.e., step 7 in Fig. 2).
B. Evaluation of Spectrum Sharing in Hyperledger Fabric In the proposed DSS system, a block of the spectrum
In this section, we evaluate the performance of spec- sharing ledger supports all transaction information, e.g., the
trum sharing in Hyperledger Fabric. As one of consortium requesting and selling information, the buying and pricing
blockchains, Fabric is a extensible blockchain system for strategy, and ID of each operator, etc. A block header of data
running distributed applications and one of the Hyperledger instance is listed, as shown in Fig. 8. We can get data hash,
projects hosted by the Linux Foundation [41]. Compared to block number, block size, and create timestamp, etc., by query-
the traditional blockchain, Fabric reaches a consensus much ing the block header. The block header only stores the data
more quickly and with less effort, and achieves higher end-to- Hash of the transaction information, while it is recorded in the
end throughput. Consequently, we leverage Fabric to evaluate block body. As listed in Table III, the data recorded in a block
the spectrum sharing assignment. gradually increases with the organization’s continual entry to
We implement the deployment of a Fabric v2.2 network the system, which results in the increase of the block size and
on 4 Linux desktops, and each that has the same configura- submission time, especially when the number of organizations
tions as illustrated in Table II. Each Linux desktop consists of is very large.
the same number of operator organizations, and organizations
4 The batch timeout defines the amount of time to wait after the first trans-
are deployed in docker containers as virtual networks. Each
action arrives for additional transactions before cutting a block, the batch
organization includes a peer node that also acts as leader, com- size is the size of transactions in a block, and the max message count is the
mitting, endorsing, and anchor peer, and an orderer node. We maximum number of messages to permit in a batch.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
12 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

Kafka, PBFT produces longer latency to reach a consen-

sus on the spectrum sharing strategy, especially when the
number of consensus nodes is larger than 16, whereas the
latencies with Raft and Kafka are less affected by the num-
ber of consensus nodes. In addition, the performance of
Kafka is better than that of Raft. That is because Kafka does
not rely on the peer-to-peer communication between nodes.
Compared with Raft and PBFT, Kafka is more suitable for
multi-operator DSS scenario, which produces less end-to-end
latency.

VII. C ONCLUSION
In this paper, we have proposed a consortium blockchain-
based secure and efficient DSS framework with guaranteed
Fig. 9. Submission time versus the number of organizations. revenue and QoS. Each operator can adaptively act as spec-
trum provider or spectrum requestor based on its demand,
TABLE III
and the spectrum allocation is made on the chain with the
B LOCK S IZE DSS smart contract. To guarantee the fairness of spectrum
sharing, we have presented a MLMF Stackelberg game to
determine the buying and pricing strategies of BOPs and
SOPs, and developed the solution leveraging KKT conditions
and RMSProp algorithm. We have implemented the DSS in
Hyperledger Fabric, and evaluated the latency and through-
put performance, which shows that the submission time and
average latency increases with the participants.

A PPENDIX A
P ROOF OF P ROPOSITION 2
Let λ1m , λ2m and λ3m be Lagrange multipliers that are asso-
ciated with the constraints in Problem 1, respectively. Then,
the Lagrangian function is given as follows:
   
Lm bm,n = Um bm,n + λ1m bm,n

N
+ λ2m Qm − pn bm,n
n=1

M
+ λ3m Bn − bm,n . (16)
m=1
The KKT conditions of Problem 1 are listed as follows.
Fig. 10. Average latency versus number of consensus nodes.
Stationarity condition:
 
∂Lm bm,n wm ηm
= − pn + λ1m
∂bm,n 1 + ηm bm,n
Then, we evaluate the submission time. As shown in Fig. 9,
the mean values of the submission time are 5.46s, 5.49s, − λ2m pn − λ3m = 0. (17)
5.70s, 6.16s and 6.73s under 10, 20, 30, 40, and 50 orga-
Primal feasibility conditions:
nizations, respectively, and they gradually increase as the
number of organizations increases. The intervals between the bm,n ≥ 0, (18)
submission time are 0.03s, 0.21s, 0.46s, and 1.32s, respec- 
N
tively. This suggests that the intervals of submission time Qm − pn bm,n ≥ 0, (19)
gradually increase with organizations continually joining the n=1
system. 
M
Finally, we compare the average latency under three clas- Bn − bm,n ≥ 0. (20)
sical consensus algorithms, i.e., practical Byzantine fault m=1
tolerance (PBFT) [42], Raft [43], and Kafka. As shown in Dual feasibility conditions:
Fig. 10, the average latency increases gradually with the num-
ber of consensus nodes. However, compared with Raft and λ1m , λ2m , λ3m ≥ 0. (21)

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 13

Complementary slackness conditions: optimal solution in Case 3, the condition with respect to pn
denoted as C3, is given by:
λ1m bm,n = 0, (22)  
Qm − N n  =n pn  bm,n  wm ηm

N
pn ≤ min , . (29)
λ2m Qm − pn bm,n = 0, (23) Bn − M m  =m bm  ,n
1 + ηm bm,n
r

n=1 max . In this case, the maxi-

Case 4: 0 < bm,n < bm,n

M
λ3m Bn − bm,n = 0. (24) mum bought bandwidth constraint and the minimum bought
m=1
bandwidth constraint are inactive. According to the condi-
tions (22), (23), and (24), we have λ1m = 0, λ2m = 0, and
According to Proposition 1, the strategy space of BOPs is λ3m = 0. Substitute them into (17), we obtain the optimal
max ], where b max > 0, i.e., b l
[0, bm,n m,n > 0 and bm,n > 0. The
r
m,n solution bm,n = wm /pn − 1/ηm .
optimal solution of the problem is given as follows considering
four different cases. A PPENDIX B
Case 1: bm,n = 0. In this case, the minimum bought P ROOF OF P ROPOSITION 3
bandwidth constraint is active. Substitute bm,n = 0 into (23)
The strategy space of SOPs is [Cn , min{Pn , P̄ }], which
and (24), we have λ2m = 0 and λ3m = 0. Combing (17), we
is nonempty, closed, and convex. Therefore, the upper-layer
have λ1m = pn − wm ηm ≤ 0. If pn < wm ηm , the con-
game exists a Nash equilibrium. As given in Proposition 2, the
dition (21) cannot be satisfied as λ1m < 0, which means
optimal solution for the buying strategy is one of four cases.
that bm,n = 0 is not feasible in this case. Otherwise, if ∗
The first and second order derivatives of bm,n with respect to
pn = wm ηm , we have λ1m = 0, and the optimal solution
pn are calculated as:
is bm,n = 0. Therefore, to achieve the optimal solution in ⎧
Case 1, the condition with respect to pn denoted as C1, is ⎪
⎪ 0, if pn ∈ C1 ∪ C3,
∗
∂bm,n ⎨ Q −N p b
given by: m n  =n n  m,n 
= − pn2
, if pn ∈ C2, (30)
∂pn ⎪ w
⎪
pn = w m η m . (25) ⎩− m , else,
p2 n

Case 2: bm,n = bm,n l . In this case, the maximum and

⎧
bought bandwidth constraint is active. As defined in (9), ⎪
⎪ 0, if pn ∈ C1 ∪ C3,
∂ 2 bm,n
∗ ⎨   
it satisfies bm,nl ≤ bm,nr , thus we have p
n ≥ (Qm − 2 Qm − N p b
n  =n n  m,n 
= , if pn ∈ C2, (31)
N
 p 
n =n n m,n b  )/(B − M
b
m =m m ,n
  ). According to (22) ∂pn2 ⎪
⎪ pn3
n
1 3
⎩ 2wm , else.
and (24), we have λm = 0 and λm = 0. Substitute p3 n
bm,n = bm,nl , λ1m = 0, and λ3m = 0 into (17), we have According to the revenue function (3), the first and second
wm ηm order derivatives of Rn with respect to pn are calculate as:
λ2m =   − 1. (26)
1 + ηm bm,n
l pn M  ∗ 
∂Rn ∗ ∂bm,n
= bm,n + (pn − Cn ) , (32)
∂pn ∂pn
If λ2m in (26) satisfies λ2m ≥ 0, i.e., pn ≤ wm ηm /(1 + m=1
ηm bm,n
l ), all of the KKT conditions are satisfied, thus the and
optimal solution is bm,n = bm,n l . Otherwise, the optimal ∗
solution does not hold in this case. Therefore, to achieve the ∂ 2 Rn M
∂bm,n ∂ 2 bm,n
∗
= 2 + (p n − C n ) . (33)
optimal solution in Case 2, the condition with respect to pn ∂pn2 ∂pn ∂pn2
m=1
denoted as C2, is given by:
As pn − Cn < pn , substituting (30) and (31) into (33), we
Qm − N have ∂ 2 Rn /∂pn2 ≤ 0. Thus, the utility function Rn of SOP is a
n  =n pn  bm,n  wm ηm
≤ pn ≤ . (27) concave function with respect to pn , and the Nash equilibrium
Bn − M
m  =m bm  ,n
1 + ηm bm,n
l
point of the upper-layer game always exists and is unique.

Case 3: bm,n = bm,nr . In this case, b r

m,n ≤ bm,n , thus we
l
R EFERENCES
N M
have pn ≤ (Qm − n  =n pn  bm,n  )/(Bn − m  =m bm  ,n ).
[1] S. Chen, Y.-C. Liang, S. Sun, S. Kang, W. Cheng, and M. Peng, “Vision,
Similarly, we have λ1m = 0 and λ2m = 0 based on (22) requirements, and technology trend of 6G: How to tackle the challenges
and (23). Substitute them into (17), we have of system coverage, capacity, user data-rate and movement speed,” IEEE
Wireless Commun., vol. 27, no. 2, pp. 218–228, Apr. 2020.
wm ηm [2] Z. Liao, C. Chen, Y. Ju, C. He, J. Jiang, and Q. Pei, “Multi-controller
λ3m = − pn . (28) deployment in SDN-enabled 6G space–air–ground integrated network,”
1 + ηm bm,n Remote Sens., vol. 14, no. 5, p. 1076, Feb. 2022.
[3] P. Kolodzy and I. Avoidance, “Spectrum policy task force,” Fed.
If λ3m in (28) satisfies λ3m ≥ 0, i.e., pn ≤ wm ηm /(1 + Commun. Comm., Washington, DC, USA, Rep. ET Docket 02-135,
ηm bm,n
r ), all of the KKT conditions are satisfied, and the Nov. 2002.
[4] P. Gu, R. Li, C. Hua, and R. Tafazolli, “Dynamic cooperative spectrum
optimal solution is bm,n = bm,n r . Otherwise, the optimal
sharing in a multi-beam LEO-GEO co-existing satellite system,” IEEE
solution does not hold in this case. Therefore, to achieve the Trans. Wirel. Commun., vol. 21, no. 2, pp. 1170–1182, Feb. 2022.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
14 IEEE TRANSACTIONS ON COGNITIVE COMMUNICATIONS AND NETWORKING, VOL. 9, NO. 1, FEBRUARY 2023

[5] Z. Li, W. Wang, J. Guo, Y. Zhu, L. Han, and Q. Wu, “Blockchain- [28] P. Gorla, V. Chamola, V. Hassija, and N. Ansari, “Blockchain based
empowered dynamic spectrum management for space-air-ground inte- framework for modeling and evaluating 5G spectrum sharing,” IEEE
grated network,” Chin. J. Electron., vol. 31, pp. 456–466, May 2022. Netw., vol. 35, no. 2, pp. 229–235, Mar./Apr. 2021.
[6] J. Park, J. G. Andrews, and R. W. Heath, “Inter-operator base station [29] L. Xue, W. Yang, W. Chen, and L. Huang, “STBC: A novel blockchain-
coordination in spectrum-shared millimeter wave cellular networks,” based spectrum trading solution,” IEEE Trans. Cogn. Commun. Netw.,
IEEE Trans. Cogn. Commun. Netw., vol. 4, no. 3, pp. 513–528, vol. 8, no. 1, pp. 13–30, Mar. 2022.
Sep. 2018. [30] Introduction of a New NR Band for LTE/NR Spectrum Sharing in Band
[7] S. Biswas, A. Bishnu, F. A. Khan, and T. Ratnarajah, “In-band full- 41/n41 (Rel-15), 3GPP Standard TS 38.331 V15.6.0, Jun. 2019.
duplex dynamic spectrum sharing in beyond 5G networks,” IEEE [31] Introduction of Enhanced Support for Dynamic Spectrum Sharing (Rel-
Commun. Mag., vol. 59, no. 7, pp. 54–60, Jul. 2021. 16), 3GPP Standard TS 38.211 V16.0.0, Dec. 2019.
[8] B. Qian, H. Zhou, T. Ma, K. Yu, Q. Yu, and X. Shen, “Multi-operator [32] S. Nakamoto. “Bitcoin: A peer-to-peer electronic cash system:
spectrum sharing for massive IoT coexisting in 5G/B5G wireless Decentralized business Review.” 2008. [Online]. Available: https://
networks,” IEEE J. Sel. Areas Commun., vol. 39, no. 3, pp. 881–895, bitcoin.org/bitcoin.pdf
Mar. 2021. [33] G. Wood et al. “Ethereum: A secure decentralised generalised transaction
[9] Y. Xu, J. Wang, Q. Wu, A. Anpalagan, and Y. Yao, “Opportunistic ledger.” Ethereum Project Yellow paper. 2014. [Online]. Available: https:
spectrum access in cognitive radio networks: Global optimization using //ethereum.github.io/yellowpaper/paper.pdf
local interaction games,” IEEE J. Sel. Top. Signal Process., vol. 6, no. 2, [34] G. O. Boateng, D. Ayepah-Mensah, D. M. Doe, A. Mohammed, G. Sun,
pp. 180–194, Apr. 2012. and G. Liu, “Blockchain-enabled resource trading and deep reinforce-
ment learning-based autonomous RAN slicing in 5G,” IEEE Trans. Netw.
[10] M. D. Mueck, V. Frascolla, and B. Badic, “Licensed shared access—
Service Manag., vol. 19, no. 1, pp. 216–227, Mar. 2022.
State-of-the-art and current challenges,” in Proc. 1st Int. Workshop Cogn.
[35] “Hyperledger architecture, volume 1: Introduction to hyper-
Cell. Syst. (CCS), 2014, pp. 1–5.
ledger business blockchain design philosophy and consensus.”
[11] M. M. Sohul, M. Yao, T. Yang, and J. H. Reed, “Spectrum access system
Hyperledger Architecture Working Group (WG). Aug. 2017. [Online].
for the citizen broadband radio service,” IEEE Commun. Mag., vol. 53,
Available: https://www.hyperledger.org/wp-content/uploads/2017/08/
no. 7, pp. 18–25, Jul. 2015.
Hyperledger_Arch_WG_Paper_1_Consensus.pdf
[12] Z. Sun, W. Liang, F. Qi, Z. Dong, and Y. Cai, “Blockchain-based [36] M. J. Osborne, An Introduction to Game Theory. New York, NY, USA:
dynamic spectrum sharing for 6G UIoT networks,” IEEE Netw., vol. 35, Oxford Univ. Press, 2004.
no. 5, pp. 143–149, Sep./Oct. 2021. [37] S. Park, “Existence theorems for generalized Nash equilibrium prob-
[13] H. Zhang, S. Leng, F. Wu, and H. Chai, “A DAG blockchain enhanced lems,” Banged Int. J. Math. Comput. Sci, vol. 1, no. 1, pp. 42–51,
user-autonomy spectrum sharing framework for 6G-enabled IoT,” IEEE Apr. 2015.
Internet Things J., vol. 9, no. 11, pp. 8012–8023, Jun. 2022. [38] F. Zou, L. Shen, Z. Jie, W. Zhang, and W. Liu, “A sufficient condi-
[14] S. Hu, Y.-C. Liang, Z. Xiong, and D. Niyato, “Blockchain and artificial tion for Convergences of adam and RMSProp,” in Proc. IEEE/CVF
intelligence for dynamic resource sharing in 6G and beyond,” IEEE IEEE Comput. Soc. Conf. Comput. Vis. Pattern Recognit. (CVPR), 2019,
Wireless Commun., vol. 28, no. 4, pp. 145–151, Aug. 2021. pp. 11127–11135.
[15] M. B. H. Weiss, K. Werbach, D. C. Sicker, and C. E. C. Bastidas, “On [39] G. Hinton, N. Srivastava, and K. Swersky, Coursera Lectures, Irvine,
the application of blockchains to spectrum management,” IEEE Trans. CA, USA. Neural Networks for Machine Learning. (2012). [Online].
Cogn. Commun. Netw., vol. 5, no. 2, pp. 193–205, Jun. 2019. Available: https://www.youtube.com/watch?v=LN0xtUuJsEI&list=
[16] “Remarks of commissioner Jessica Rosenworcel at mobile world PLoRl3Ht4JOcdU872GhiYWf6jwrk_SNhz9&index=42
congress Americas,” Fed. Commun. Comm., Los Angeles, CA, USA, [40] N. Chen, N. Gravin, and P. Lu, “On the approximability of budget
document DOC-354091A1, Sep. 2018. [Online]. Available: https://docs. feasible mechanisms,” in Proc. 22nd Annu. ACM-SIAM Symp. Discr.
fcc.gov/public/attachments/DOC-354091A1.pdf Algorithms, 2011, pp. 685–699.
[17] H. Yao, T. Mai, J. Wang, Z. Ji, C. Jiang, and Y. Qian, “Resource trading [41] E. Androulaki et al., “Hyperledger fabric: A distributed operating system
in blockchain-based Industrial Internet of Things,” IEEE Trans. Ind. for permissioned blockchains,” in Proc. 13th Eur. Conf. Comput. Syst.,
Informat., vol. 15, no. 6, pp. 3602–3609, Jun. 2019. 2018, pp. 1–15.
[18] J. Ye, X. Kang, Y.-C. Liang, and S. Sun, “A trust-centric privacy- [42] M. Castro and B. Liskov, “Practical Byzantine fault tolerance,” in Proc.
preserving blockchain for dynamic spectrum management in IoT 3rd Symp. Oper. Syst. Des. Implement., 1999, pp. 173–186.
networks,” IEEE Internet Things J., vol. 9, no. 15, pp. 13263–13278, [43] D. Ongaro and J. Ousterhout, “In search of an understandable consensus
Aug. 2022. algorithm,” in Proc. USENIX Annu. Tech. Conf., Philadelphia, PA, USA,
[19] H. Zhang, S. Leng, and H. Chai, “A blockchain enhanced dynamic spec- 2014, pp. 305–319.
trum sharing model based on proof-of-strategy,” in Proc. IEEE Int. Conf.
Commun. (ICC), 2020, pp. 1–6. Zuguang Li received the B.S. degree in elec-
[20] F. Wilhelmi and L. Giupponi, “On the performance of blockchain- tronic information science and technology from the
enabled RAN-as-a-service in beyond 5G networks,” in Proc. IEEE Nanjing University of Aeronautics and Astronautics,
Global Commun. Conf. (GLOBECOM), 2021, pp. 1–6. Nanjing, China, in 2020, where he is currently
[21] Y. Xiao et al., “Decentralized spectrum access system: Vision, chal- pursuing the M.S. degree in communication and
lenges, and a blockchain solution,” IEEE Wireless Commun., vol. 29, information system. His research interests mainly
no. 1, pp. 220–228, Feb. 2022. focus on blockchain technologies and dynamic spec-
trum sharing.
[22] R. Zhu, H. Liu, L. Liu, X. Liu, W. Hu, and B. Yuan, “A blockchain-
based two-stage secure spectrum intelligent sensing and sharing auction
mechanism,” IEEE Trans. Ind. Informat., vol. 18, no. 4, pp. 2773–2783,
Aug. 2022.
Wei Wang (Member, IEEE) received the B.Eng.
[23] T. Maksymyuk et al., “Blockchain-empowered framework for degree in information countermeasure technology
Decentralized network management in 6G,” IEEE Commun. Mag., and the M.Eng. degree in signal and information pro-
vol. 58, no. 9, pp. 86–92, Sep. 2020. cessing from Xidian University in 2011 and 2014,
[24] Z. Li, W. Wang, J. Guo, Y. Zhu, L. Han, and Q. Wu, “Blockchain- respectively, and the Ph.D. degree in electrical and
assisted dynamic spectrum sharing in the CBRS band,” in Proc. electronic engineering from Nanyang Technological
IEEE/CIC Int. Conf. Commun. China (ICCC), 2021, pp. 864–869. University, Singapore, in 2018. From September
[25] A. Khanna, P. Rani, T. H. Sheikh, D. Gupta, V. Kansal, and J. J. P. C. 2018 to August 2019, he was a Postdoctoral Fellow
Rodrigues, “Blockchain-based security enhancement and spectrum sens- with the Department of Electrical and Computer
ing in cognitive radio network,” Wireless Personal Commun., Jul. 2021. Engineering, University of Waterloo, Canada. He is
[26] S. Zheng, T. Han, Y. Jiang, and X. Ge, “Smart contract-based spec- currently a Professor with the Nanjing University
trum sharing transactions for multi-operators wireless communication of Aeronautics and Astronautics. His research interests include wireless
networks,” IEEE Access, vol. 8, pp. 88547–88557, 2020. communications, space-air-ground integrated networks, wireless security, and
[27] Z. Zhou, X. Chen, Y. Zhang, and S. Mumtaz, “Blockchain-empowered blockchain. He was awarded the IEEE Student Travel Grants for IEEE
secure spectrum sharing for 5G heterogeneous networks,” IEEE Netw., ICC 2017 and the Chinese Government Award for outstanding self-financed
vol. 34, no. 1, pp. 24–31, Jan./Feb. 2020. students abroad.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.
LI et al.: MULTI-OPERATOR DSS FOR WIRELESS COMMUNICATIONS 15

Qihui Wu (Senior Member, IEEE) received the Xianbin Wang (Fellow, IEEE) received the Ph.D.
B.S. degree in communications engineering, and degree in electrical and computer engineering from
the M.S. and Ph.D. degrees in communications the National University of Singapore in 2001.
and information systems from the Institute of He is a Professor and the Tier-1 Canada Research
Communications Engineering, Nanjing, China, in Chair with Western University, Canada. Prior to join-
1994, 1997, and 2000, respectively. From 2003 to ing Western, he was with Communications Research
2005, he was a Postdoctoral Research Associate with Centre Canada (CRC) as a Research Scientist/Senior
Southeast University, Nanjing. From 2005 to 2007, Research Scientist from July 2002 to December
he was an Associate Professor with the Institute 2007. From January 2001 to July 2002, he was
of Communications Engineering, PLA University of a System Designer with STMicroelectronics. He
Science and Technology, Nanjing, where he is cur- has over 450 highly cited journal and conference
rently a Full Professor. From March 2011 to September 2011, he was an papers, in addition to 30 granted and pending patents and several stan-
Advanced Visiting Scholar with the Stevens Institute of Technology, Hoboken, dard contributions. His current research interests include 5G/6G technologies,
USA. Since 2016, he has been with the Nanjing University of Aeronautics Internet-of-Things, communications security, machine learning and intelligent
and Astronautics and has been appointed as a Distinguished Professor. He was communications.
appointed the Changjiang Distinguished Professorship in 2016. His current Dr. Wang has received many awards and recognitions, including the Canada
research interests span the areas of wireless communications and statistical Research Chair, the CRC President’s Excellence Award, the Canadian Federal
signal processing, with emphasis on system design of software defined radio, Government Public Service Award, the Ontario Early Researcher Award, and
cognitive radio, and smart radio. the six IEEE Best Paper Awards. He currently serves/has served as an Editor-
in-Chief, an Associate Editor-in-Chief, an Editor/Associate Editor for over 10
journals. He was involved in many IEEE conferences, including GLOBECOM,
ICC, VTC, PIMRC, WCNC, CCECE, and CWIT, in different roles, such as
the General Chair, the Symposium Chair, a Tutorial Instructor, the Track Chair,
the Session Chair, the TPC Co-Chair, and a Keynote Speaker. He has been
nominated as an IEEE Distinguished Lecturer several times during the last
ten years. He is currently serving as the Chair of IEEE London Section and
the Chair of ComSoc Signal Processing and Computing for Communications
Technical Committee. He is a Fellow of Canadian Academy of Engineering,
a Fellow of Engineering Institute of Canada, and an IEEE Distinguished
Lecturer.

Authorized licensed use limited to: North West University. Downloaded on April 20,2023 at 14:51:27 UTC from IEEE Xplore. Restrictions apply.