Adaptive neuro-fuzzy behavioral learning strategy for effective decision making in the fuzzy-based cloud service negotiation framework

Abstract

Future cloud computing creates a new trend of opting service over the internet through some intelligent third-party broker. In cloud market, both consumer and provider compete with each other against the conflicting requirements, and the competition among cloud providers to trade their services to potential consumers of cloud market. There is an increasing need for automated negotiation framework to quickly reach agreement in competitive cloud market which can provide maximum utility value and success rate among the negotiating parties. Researchers develop various behavioral learning negotiation strategies (such as market driven) in the existing negotiation frameworks for maximizing either the choice of utility value or success rate of parties. Moreover these strategies can be applicable to the environment, where the opponent’s behaviors are predictable or precisely known. It may be daunting to apply in the dynamically varying competitive cloud market. So, the proposed Adaptive Neuro-Fuzzy Behavioral Learning (ANFBL) strategy can be applicable, where the opponent’s behavior is partially and imprecisely known. Therefore, the proposed strategy can maximize both utility value and success rate without compromising either choice. An extensive simulation is conducted to evaluate the efficiency of strategies which shows that proposed strategy achieve higher utility and higher success rate than existing learning approach, without any negotiation conflict among the parties.

Keywords

Cloud computing automated negotiation framework fuzzy behavioral learning adaptive neuro-fuzzy

1. Introduction

In this decade, cloud computing is being used in different application domains such as industry, government, and science [1]. The major vision of cloud computing is to offers huge computing power and information technology service delivery to consumers, which can brings the similar comfort of traditional utilities like electricity, water, and gas [2]. In recent years, cloud computing technology has become an essential part of the enterprise information technology business for its on-demand self service provisioning and metering of services. It offers high computing resource availability and throughput for the services after confirming the SLA agreement between the participants (consumer and service provider) [3]. The SLA is a contract made between the participants to promise the Quality of Service as goals for the vision of cloud computing [4]. The Quality of Service goals and economic conditions such as pricing and violation terms to be established are clearly stated in the SLA. This SLA is classified into static (provider predefined) SLA and dynamic (negotiated) SLA [5]. In the former case, a generic model of the SLA template is provided to all the consumers. In the latter case, the consumer and provider undergo a series of negotiation processes between them to accomplish a mutually agreed SLA template.

In general, cloud providers like Amazon EC2, and Microsoft Azure define a general SLA document for all consumers that assure to guarantee 99.9% service availability. This type of static SLA is automatically recognized as soon as consumer confirms the service request through online credit card payment. The provisioning of services through the static SLA is a semi-customized service provisioning mechanism which is in-compliance due to predefined non-functional properties of Quality of Service goals specified by the providers. Current cloud management systems provide a traditional SLA-based service provisioning and metering mechanism [6, 7]. SLA refers to an agreement containing set of functional and non-functional properties of the service mutually agreed between the service consumer and the service provider [8]. Only semi-customized service provisioning with static SLA is available in today’s cloud management system. Forecasts indicate that a consumer with specialized Quality of Service requirements may not be satisfied by the provider for maximizing their revenue. In addition, the provider cannot provide differentiated service provisioning to consumers with specialized quality expectations.

To satisfy such future demands, an SLA-oriented cloud management system mechanism is proposed in the previous research work [9]. It highlights the demand of negotiation framework establishment in the service layer of cloud (available on top of application layer). In futuristic cloud computing vision, the negotiation framework is addressed as an imperative architectural component of SLA-oriented resource management [10, 11]. This scenario motivates the research work towards the development of negotiation framework which can satisfy the customized cloud service provisioning mechanism. In addition, the cloud technology accomplishes the service discovery, scaling, monitoring, and decommissioning operations in dynamic manner, there is a demand for automated cloud service negotiation between consumer and provider [12].

This cloud-based resource provisioning will dynamically configure the virtual machine, which is in need of an SLA negotiation between the service consumer and service provider, using an intermediate third party broker. The cloud agency can act as a broker for maintaining dynamic provisioning, monitoring, and reconfiguration of the cloud resources on behalf of the user [13]. Since, the cloud service negotiation participants such as service consumer, broker, and service provider are independent bodies with different requirements, policies, and objectives; there is a need for a negotiation framework among the participants to resolve their differences [14]. In a real-time e-commerce negotiation problem, the broker-based negotiation framework using other computing paradigms, like the grid and cluster, was restricted to resource constraints due to its negotiation complexity [15, 16]. Therefore, to overcome the resource constraint, a negotiation framework needs to be presented over the cloud environment, due to its elastic scaling of resources. Negotiation framework is the composition of various automated components like negotiation strategy, negotiation protocol, automated agent based communication among the negotiating parties as defined in the next section.

2. Introduction

In this decade, cloud computing is being used in different application domains such as industry, government, and science [1].

2.1. Negotiation framework

Initially, the negotiation concept was used for reserving bandwidth in the SLA-oriented provisioning architecture for making it a Quality of Service-aware internet [17]. Later, end-to-end Quality of Service contract negotiation was introduced among network service providers using the distributed reinforcement [18] and Q-learning algorithm for their long-term benefits. The Cloud Computing Background Key Exchange [19] and HiTrust negotiation model is proposed to provide a secure scheduling and trust relationship among the distributed service, respectively. In the communication perspective, the negotiation framework can be classified into direct (straight forward) negotiation and indirect (intermediate third party) negotiation. The direct type is preferred for single market (simple) negotiation with two parties (service consumer and service provider), whereas an indirect model is used for multiple inter-related market (complex) negotiation with three parties (service consumer, broker, and service provider) [20]. In direct negotiation, there is a point-to-point negotiation process between the service provider and service consumer. In case of indirect negotiation, a third-party is used for the negotiation process via a single or more than one trusted third-party broker.

2.2. Negotiation strategy

The negotiation strategy describes the negotiation process among participants using a standard communication protocol and the operational behavior of the offer or counter-offer generated by the participant’s algorithm at the negotiating end. A process of offer or counter offer generation, selection, and evaluation algorithm constitutes the negotiation strategy of the participants [21]. The participants may attempt to generate attractive offers in order to influence their respective counterpart by providing a dynamic concession in offers without receiving a concession in counter offers. Most research work follows the trade-off and concession-making algorithm for the negotiation strategy [22, 23]. A novel time constrained SLA negotiation strategy is proposed to cope up with the limited time duration of the negotiation session [24]. In the negotiation strategy, concession is given using the time, opportunity, and competition function. Here, the time function follows the conservative, conciliatory, and linear concession-making strategy [25]. To support negotiation with incomplete information, a co-evolutionary learning method is used in the negotiation strategy with two types of Estimation Distribution Algorithms (EDAs) such as conventional EDAs and novel Improved Dynamic Diversity Controlling EDAs [26].

The automated negotiation strategy can be developed without learning ability or with learning ability depending on the environment [27]. Much research works are available in the former case of negotiation strategy without learning ability where the negotiation attributes like price, time-slot, and speed are constantly changes over the time [28 –31]. In the later case of negotiation strategy with learning ability, very few research works are presented using behavioral learning approaches like Bayesian learning [32], neural network learning [33, 34], bulk negotiation behavioral learning [35], adaptive probabilistic behavior learning [36], Reinforcement learning [37 –39], Evolutionary behavior learning [40], Co-evolutionary learning [41]. Moreover these strategies can be applicable to the environment, where the opponent’s behaviors are predictable or precisely known. Hence, there is lack of negotiation strategy with learning ability, where the opponent’s behavior is partially and imprecisely known (such as uncertain information about the preferences of opponent agent).

To the best of our knowledge, only very few research works are available in the context of fuzzy based negotiation strategy. An agent-based fuzzy constraint directed negotiation model is developed for solving the problem of supply chain scheduling and planning without using any third party agent [42, 43]. The same fuzzy based negotiation mechanism is extended for establishing the SLA between the consumer and provider in cloud computing environment [44]. To enhance the negotiation outcome (satisfaction level) of consumer and buyer, an automated negotiation process is utilized with fuzzy inference system for generating offer and counter-offer based on the requirements and preferences of negotiating parties [45]. To further improve the negotiation outcome, evolutionary approach is used in the negotiation strategy for learning the new relaxed criteria fuzzy rules [46]. An artificial intelligence approach is embedded into negotiation process to form an adaptive fuzzy logic strategy for learning the behavior of the opponent agent [47]. These strategies lack of third party broker involvement and also lack of opponents fuzzy behavior learning mechanism during negotiation process. So, this research work focuses on developing broker based adaptive neuro-fuzzy behavioral learning strategy for enforcing behavioral learning concept under uncertain information. Moreover, this research work includes the non functional parameters like service name attribute, service requirement attribute, service negotiation parameter attribute, and functional parameters like total negotiation time, communication overhead, utility value, and success rate.

2.3. Negotiation protocol

The traditional model of negotiation reaches the agreement through the exchange of offer and counter offer between the parties using Contract Net Protocol and Alternate Offer Protocol. In the Contract Net Protocol [48], one end of the negotiation party can send the offer and the opponent party can either accept or reject the offer. The actual concept of negotiation is not realized in this protocol, because the opponent is not sending any counter offer. So, to realize the negotiation concept, an alternate offer protocol [49] is proposed where the opponent can accept, reject, and modify the offer for sending the counter offer. Later, Service Negotiation and Acquisition Protocols were developed in the grid environment for supporting the reliable management of remote SLAs [50]. In order to establish the complex multi-party and composite agreement, a requirement-driven negotiation protocol is developed by iteratively discovering the dependencies of the opponent’s requirements and offers [51]. An automated negotiation is developed in dynamic environments under restricted negotiation time and interdependencies with three protocols: Combinatorial Action Protocol, Cluster Bidding Protocol, and Mediated Negotiation Protocol [52]. To automate the cloud service composition, a Focused Selection Contract Net Protocol is used for dynamically selecting the list of cloud agents and their services [53]. A machine learning concept is integrated in Wrap-5 cross-layer routing protocol [54]. According to recent literature study [55] and past literature study [56], an Alternate Offer Protocol is more popularly used in all the real time applications due to its significant realization and modeling of negotiation process during the implementation.

3. Proposed fuzzy-based cloud service negotiation framework

A conceptual architecture of the proposed fuzzy-based cloud service negotiation framework is represented in Fig. 1. The major components of this framework includes service consumer, intelligent third party broker, service provider, directory facilitator registry service, Jade gateway agent service, universal description discovery and integration registry service, and adaptive neuro-fuzzy behavior learning strategy. To automate the negotiation process, agent based technology is embedded to mimic the behavior of all the above components. Here, the Service Consumer Agent (SCA), Intelligent Third-Party Broker Agent (ITBA), and Service Provider Agent (SPA) are involved in negotiating the service on behalf of service consumer, intelligent third party broker, and service provider respectively. The SCA generates the requirements and preferences of service consumer and forward the same to intelligent third party broker. This component will identify the appropriate ITBA to negotiate the service with multiple SPAs. Now, the actual negotiation process will takes place between the ITBA and SPAs and finally suggest the committed service to SCA. During the automation of negotiation process, the sequence of offer and counter offer generated by the ITBA and SPA will exploits certain negotiation strategy at each negotiation states. In the proposed framework, the ITBA exploits the adaptive neuro-fuzzy behavioral learning strategy for maximizing the utility value and success rate among the negotiation parties. This strategy will learn the behavior of the opponent at each negotiation state and generates the adaptive counter-offer. Also, it increases the negotiation coordination among the negotiating parties without causing any negotiation conflict with the SPAs. Coordination may leads to commitment of SLA and conflict may leads to negotiation break off among the negotiating parties. Thereby, behavior learning will maximize the utility value and the commitment of SLA will maximize success rate.

Fig.1

Fuzzy-based cloud service negotiation framework.

3.1. Adaptive Neuro-Fuzzy Behavioral Learning Strategy

The conceptual architectural of adaptive neuro-fuzzy behavioral learning strategy consist of several components like offer sensor agent, control initiator, sensorial data administrator, fuzzy behavioral learning system, negotiation controller, basic negotiation behaviors, and counter-offer actuator agent as shown in Fig. 2. All ITBA will invoke this strategy after receiving the negotiation offer from the opponents for the sake of generating appropriate counter offer to the SPAs. An offer sensor agent always senses the offer and gives perceived negotiation state information to the sensorial data administrator and control initiator for further analysis. A control initiator component supplies the precision value of several binary predicates (or switching of agent’s mission) based on the computations of sensorial observations from offer sensor agent. Next, sensorial data administrator start aggregating the negotiation offers received from sensors and matches with approximate behavioral rules available in the component of basic negotiation behaviors. Later, this matching behavior rules with different activation levels has to follow by the ITBA during the process of counter offer generation. Additionally, the sensorial data administrator calculates precision values of the predicates exploited by the rule-matching algorithm. Based on mission switching and matching behavioral rules information, the fuzzy behavior learning system will implements the reinforcement learning capability by continuously updating the rule base consisting of fuzzy context rules (meta-rules). It operates on the inhabitants of fuzzy rules that characterize the antecedent (behavioral context) and the consequent (basic behavior) of the rule.

Fig.2

Architecture of adaptive neuro-fuzzy behavioral learning model.

At each negotiation state, this system helps in improving the speed of negotiation convergence by learning best behavioral context for any course of basic behavior employed in opponent’s agent. Moreover, the reinforcement learning evaluates the performance among the agents at the end of each negotiation state and distributes the reinforcement values to the rules that control the agent during negotiation process. Each rule is estimating the strength of the antecedent to represent the appropriate context for the application of basic behavioral activation by the consequent. According to sensed negotiation and current mission switching initiated by the control initiator, a negotiation controller component computes the activation levels of all behavioral rules which are to be incorporated during the negotiation states. Finally, it choose the appropriate levels of behavior activation in the basic negotiation behaviors as a collection of operating functions with the intention of operating the agent behavior in parallel.

3.2. Formulation of fuzzy based negotiation process using reinforcement learning and artificial neural networks

The actual bilateral negotiation process between the ITBA and SPA can be formulated according to the adaptive neuro-fuzzy behavioral learning strategy using reinforcement learning and artificial neural network techniques. During the negotiation process, sequence of offer ρ_a→b and counter offer ρ_b→a are generated by the broker agent a∈ { ITBA₁, ITBA₁, … ITBA_n } and provider agent b∈ { SPA₁, SPA₁, …, SPA_m } with respect to multiple negotiation attribute or issue $i \in {i_{1}^{Price}, i_{2}^{Time - slot}, i_{3}^{Quality}, \dots i_{n}^{Quality}}$ . Here, the broker agents adopt the proposed behavioral learning strategy for generating the sequence of offers. Similarly, the provider agents also adopt any one of the negotiation strategy like trade-off, linear, conciliatory and conservative concession for generating the respective counteroffer. At any time stamp t_k∈ { t₁, t₂, …, t_n }, broker agent a’s negotiation offer decision on provider agent b’s past counter-offer at time t_k-1 is defined in Equation (1). $ρ_{a \to b}^{t_{k}} = {\begin{matrix} accept, if U (ρ_{b \to a}^{t_{k - 1}}) \geq U (ρ_{a \to b}^{t_{k - 2}}) \\ reject, if U (ρ_{b \to a}^{t_{k - 1}}) < U (ρ_{a \to b}^{t_{k - 2}}) \\ offer, otherwise \end{matrix}$ (1)

Where, the total utility of the offer $U (ρ_{a \to b}^{t_{k - 2}})$ or counter offer $U (ρ_{b \to a}^{t_{k - 1}})$ is computed with respect to number of attributes involved in negotiation process as shown in Equation (2). In order to maximize the utility, negotiator can give different preferences over the negotiation attribute by assigning appropriate weight to the attributes. $U (ρ_{a \to b}^{t_{k - 2}}) = \sum_{p = 1}^{n} W_{p} \times u (ρ_{a \to b}^{t_{k - 2}} [i_{p}]$ (2)

Let $u (ρ_{a \to b}^{t_{k - 2}} [i_{p}])$ denotes the utility value of the attribute i_p over offer $ρ_{a \to b}^{t_{k - 2}}$ , W_p denotes the weight assigned to the respective negotiation attribute i_p, weight of each attribute range 0 < W_p < 1 and thereby the total weight of attribute is equal to one (W₁ + W₂ + … + W_n = 1). A value of each attribute range is [0,1] and it can be defined in terms of fuzzy values as [i_pVL, i_pL, i_pM, i_pH, i_pVH]. Here, the offer attribute value VL, L, M, H, and VH stands for very low, low, medium, high, and very high range of values respectively. Normally, the highest utility value of one participant offer $u (ρ_{a \to b}^{t_{k}} [i_{pVH}^{ITBA}]) = 1$ may give lowest utility value for the opponent $u (ρ_{a \to b}^{t_{k}} [i_{pVH}^{SPA}]) = 0$ . Similarly, the lowest utility value of same participant offer $u (ρ_{a \to b}^{t_{k}} [i_{pVL}^{ITBA}]) = 0$ may give highest utility value for the opponent $u (ρ_{a \to b}^{t_{k}} [i_{pVL}^{SPA}]) = 1$ .

During the bilateral negotiation process, the ITBA only initiates the negotiation process by generating the offer proposal and also receives the corresponding counter offer proposal from the SPA. From the received offer, broker agent extracts the values of negotiation attributes for computing the value of opponent’s preference and concession degree by considering the negotiation zone of the participants. Negotiation zone $i_{p \in (1, n)}^{ZONE}$ is defined as the range of negotiation attributes value between broker agent and provider agent initial values, and agreement zone is defined as the range of negotiation attributes value between broker agent and provider agent reservation values. Here, the initial value and reservation value represents the minimum and maximum value agreed to pay by the negotiating parties.

To effectively reach the agreement among the negotiating parties, a concession degree and preference degree member functions are defined as shown in Equations (3 and 4). This member function is used to predict the opponent’s preference degree (ω_PD) and concession degree (ω_CD) at each negotiation state. Predicted values will be helpful during the subsequent negotiation states for making appropriate fuzzy behavioral decision by the proposed strategy. $ω_{PD} = \frac{i_{1}^{CR} + i_{2}^{CR} + \dots + i_{n}^{CR}}{i_{1}^{ZONE} + i_{2}^{ZONE} + \dots + i_{n}^{ZONE}}$ (3) $ω_{CD} = \frac{i_{1}^{PR} + i_{2}^{PR} + \dots i_{n}^{PR}}{i_{1}^{R} + i_{2}^{R} + \dots + i_{n}^{R}}$ (4)

The joint membership (desirability) function d_{n∈{1,2,3,4}} can be expressed between two membership function f_k∈{1,2} (ω_PD) and f_k∈{1,2} (ω_CD) as specified in Equation (5). $d_{n} = min [f_{k \in {1, 2}} (ω_{PD}), f_{k \in {1, 2}} (ω_{CD})]$ (5)

Each d_n impacts more on to the concession rate derived from fuzzy rule FR_{f(ω_PD),f(ω_CD)}, which determines the different pair of membership functions as states in Table 1. By exploiting these membership functions, one negotiating party can evaluate the behavior of opponent and then determine the appropriate concession decision during counter offer generation process to quickly reach an agreement. In order to handle dynamic situations of negotiation process, 25 different fuzzy rules adopts various concession rate according to five levels of membership functions. An average concession rate ω_R can be defined by means of center of area method for the defuzzification purpose as specified in Equation (6).

Table 1

Fuzzy rule generation with respect to concession and preference degree

f_k∈{1,2} (ω_PD)	f_k∈{1,2} (ω_CD)
	VL	L	M	H	VH
VL	FR _VL,VL	FR _VL,L	FR _VL,M	FR _VL,M	FR _VL,VH
L	FR _M,VL	FR _L,L	FR _L,M	FR _L,H	FR _L,VH
M	FR _M,VL	FR _M,L	FR _M,M	FR _M,H	FR _M,VH
H	FR _H,VL	FR _H,L	FR _H,M	FR _H,H	FR _H,VH
VH	FR _VH,VL	FR _VH,L	FR _VH,M	FR _VH,H	FR _VH,VH

$\bar{ω_{R}} = \frac{\sum_{n} d_{n} \times (ω_{R})_{n}}{\sum_{n} d_{n}}$ (6)

At time t_k the broker agent can generate the counter offer attributes $ρ_{a \to b}^{t_{k}}$ based on received offer $ρ_{b \to a}^{t_{k - 1}} [i_{p}^{SPA}]$ at time t_k-1 by adding or removing the attributes range $(i_{1}^{R} + i_{2}^{R} + \dots + i_{n}^{R})$ with average concession rate as defined in Equation (7). $(i_{p}^{ITBA})^{t_{k}} = (i_{p}^{SPA})^{t_{k - 1}} \oplus i_{p}^{R} \times \bar{ω_{R}}$ (7)

Let the operator ⊕ denotes + and – for ITBA and SPA respectively. The proposed negotiation strategy generates the counteroffer by exploiting triangular shape to symbolize the fuzzy based preference and concession degree membership functions with respect to five levels as stated in Figs. 3 and 4 respectively.

Fig.3

Preference degree membership function.

Fig.4

Concession degree membership function.

In addition, the proposed negotiation framework defines the fuzzy behavior as triplet <C, D, E>. Let C be the negotiation context of application behavior that describes the applicability of the behavior with respect to situations. D denotes the desirability function computed at each negotiation state-action pair <s, a>that describes the desirability of performing action a at the negotiation state s of agent to realize the basic ability. Finally, E represents the element or object of negotiation environment on which the basic ability is applied. Enabling a fine composition of desirability function D can be implemented with respect to different behavioral activation at the same time, by means of fuzzy rules as given in Equation (8). $\begin{matrix} If (negotitation state IS S_{i}) \\ THEN (action IS A_{i}) \end{matrix}$ (8)

Where each rule can be related with a function in terms of T-norms as defined in Equation (9). The desirability function of the negotiation behavior is defined in terms of T-conorms as shown in Equation (10). $D_{S_{i} \to A_{i}} (s, a) = μ_{S_{i}} (s) \otimes μ_{A_{i}} (a)$ (9) $D_{B} = \oplus_{i = 1, 2, \dots, n} D_{S_{i} \to A_{i}} (s, a)$ (10)

Let n be the number of fuzzy rules that implements the basic behavior. The negotiation context of application can be defined in terms of logical combination of predicates such as global context and environmental context. Enforcing the negotiation context in behavioral activation is realized by the fuzzy rule as shown in Equation (11). $\begin{matrix} If (negotiation state s IS in context C) \\ THEN (apply behavior A) \end{matrix}$ (11)

To make more evidence, distinguish global and environmental context by rewriting the above rule as defined in Equation (12).

$\begin{matrix} If (\begin{matrix} negotiation state s IS in \\ environmental context EC \end{matrix}) \\ THEN (apply behavior A) \\ WHEN (\begin{matrix} negotiation state s IS in \\ global context GC \end{matrix}) \end{matrix}$ (12)

Here the applicability conditions in the rule level are treated as mission level equivalent of global context. Therefore, replace the above rule as shown in Equation (13). $\begin{matrix} If (\begin{matrix} negotiation state s IS in \\ environmental context EC \end{matrix}) \\ THEN (apply behavior A) \\ When (mission is M) \end{matrix}$ (13)

Let M denotes the mission which occurs when the state s appears in the global context GC.

The mission M is formally defined as 4-touple <G, AC, E, CS>. Where G is the goal of mission that allows the agent to accomplish the task, AC denotes the applicability conditions in which mission have to be carried out, E is the element reference to the developed mission, and CS denotes the coordination of strategy among agents simple behavior that recognizes the mission. The basic idea of proposed adaptive neuro-fuzzy learning strategy is to formulate the model by applying the reinforcement learning capability in artificial neural network in order to adapt the negotiation parameters and features of fuzzy logic system. To further have the transparency, neural network model make use of rule-based fuzzy reasoning during its construction. A major advantage of this model is to tune the rules of fuzzy behavioral learning system using appropriate behavioral decision and learning algorithm exploited in neural networks as shown in Fig. 5. The proposed fuzzy behavioral learning system consist of three layer feed-forward model with input layer (crisp input), hidden layer (rule) and output layer (crisp output). An input layer deals with negotiation offers received by the broker agents during the negotiation process. Hidden layer deals with set of fuzzy if-then rules stated in Table 1. Finally, the output layer represents the action aiming at generating the negotiation counteroffer decision at the broker agent.

Fig.5

Neuro-fuzzy behavioral decision.

4. Experimental evaluation

The JADE simulation tool [57] is used to create the experimental setup with set of negotiation participants like SCA, SPA, and ITBA. Here, the real scenario of this experiment setup is to simulate the e-commerce cloud service negotiation happening in the multi cloud environment. Usually the actual bilateral negotiation process will takes place between the ITBA and SPA. So, the preferences of these agents are given in the front-end portal to start the actual negotiation process as depicted in Fig. 6. A bilateral negotiation process between the agents can be visualized through the sniffer agent as shown in Fig. 7. In order to evaluate the proposed adaptive neuro-fuzzy behavioral learning strategy against the existing strategy, this research work considered the parameters such as utility value and success rate of the negotiating participants. For effective validation, benchmark dataset used in the previous research study (Rajavel, 2016) [35, 36] is considered as shown in Table 2. After the simulation, the normalized utility value and success rate are observed by varying the number of negotiation rounds with respect to different negotiation strategies as represented in Table 3.

Fig.6

Negotiation preferences of ITBA and SPA.

Fig.7

Visualization of bilateral negotiation process using sniffer agent.

Table 2

Experimental setting of e-commerce cloud service negotiation process

Input Attributes	Settings
Broker Agent	Provider Agent
Initial Price	[10,60]	[200,250]
Reserved Price	[200,250]	[10,60]
Initial Time-slot	[10,60]	[300,350]
Reserved Time-slot	[300,350]	[10,60]
Negotiation Deadline	[50,200]	[50,200]
	Rounds	Rounds
Negotiation Agent	[5,20, 5,20]	[5,20, 5,20]

Table 3

Performance of negotiation strategies

ITBA Vs.SPA Strategies	50 Rounds		100 Rounds		200 Rounds		500 Rounds
	Utility value	Success rate	Utility value	Success rate	Utility value	Success rate	Utility value	Success rate
Conciliatory vs. conciliatory	0.049	0	0.098	0	0.196	1	0.490	1
Conservative vs. conciliatory	0.049	0	0.426	1	0.426	1	0.426	1
ANFBL vs.conciliatory	0.919	1	1	1	1	1	1	1
Conciliatory vs. conservative	0.491	0	0.983	1	1	1	1	1
Conservative vs. conservative	0.491	1	0.557	1	0.557	1	0.557	1
ANFBL VS. conservative	0.919	1	1	1	1	1	1	1

It is clear from the comparison table, that the performance of proposed adaptive neuro-fuzzy behavioral learning strategy outperforms the existing conciliatory and conservative strategies in terms of utility value and success rate with respect to 50 negotiation rounds. After increasing the negotiation rounds to 100, 200 and 500, the performance of proposed strategy further increases to maximum extent due to the enforcement of a fuzzy behavioral learning system model in the strategy.

The major advantage of the proposed ANFBL strategy will give maximum utility value and success rate during negotiation process without any break-off among the negotiating participants. Only for minimum 50 negotiation round the success rate is average, after increasing the negotiation round more than 100 will always gives the maximum level of success rate. This achievement is possible due to fuzzy based behavioral learning approach exploited in the negotiation strategy. Whereas in the existing strategy various approach like conciliatory and conservative approach is used, which created constant rate of concession to negotiation offer or counter offer. But, in the proposed ANFBL strategy concession for the offer or counter offer is generated according to the behavior of the opponent’s. Thereby, the proposed ANFBL strategy can provide more utility value and success rate during the negotiation process without any bothering of opponent’s negotiation behavior.

5. Conclusion and feature work

The proposed research work evaluates the performance of negotiation frameworks with respect to different negotiation strategies. This reach work identifies that the proposed adaptive neuro-fuzzy behavioral learning strategy outperforms the existing conciliatory and conservative strategies in terms of utility value and success rate at the minimum level of negotiation rounds. After increasing to maximum level of negotiation rounds, the proposed strategy always provides the utility value and success rate due to learning capability embedded in the strategy. To further increase the performance of the negotiation framework, the degree of trust worthiness and risk membership functions can be added in the fuzzy behavioral learning mechanism. The validity of the proposed negotiation framework with respect to different combination of negotiation strategies are demonstrated through exhaustive simulation experiments. In future, this negotiation strategy can also be extended with cognition and evidence based behavioral learning approaches for further maximization of utility value and success rate by minimizing negotiation conflict among the participants.

References

A.G.

Garcia ,

I.B.

Espert and

V.H.

Garcia , SLA-driven dynamic cloud resource management, Future Generation Computer Systems 31 (2014), 1–11.

Buyya ,

C.S.

Yeo ,

Venugopal ,

Broberg and

Brandic , Cloud computing and emerging IT platforms: Vision, hype, and reality for delivering computing as the 5th utility, Future Generation Computer Systems 25 (2009), 599–616.

V.C.

Emeakaroha ,

Brandic ,

Maurer and

Dustdar , Cloud resource provisioning and SLA enforcement via LoM2HiS framework, Concurrency and Computation: Practice and Experience 25 (2013), 1462–1481.

Maurer ,

Brandic and

Sakellariou , Adaptive resource configuration for Cloud infrastructure management, Future Generation Computer Systems 29(2013),472–487.

Macias and

Guitart , SLA negotiation and enforcement policies for revenue maximization and client classification in cloud providers, Future Generation Computer Systems 41 (2014), 19–31.

Rajavel and

Thangarathanam , Optimizing negotiation conflict in the cloud service negotiation framework using probabilistic decision making model, The Scientific World Journal 2015 (2015), 1–16.

Rajavel and

Thangarathanam , A negotiation framework for the cloud management system using similarity and gale shapely stable matching approach, KSII Transactions on Internet and Information Systems 9 (2015), 2050–2077.

B.P.

Rimal and

Choi , A service-oriented taxonomical spectrum, cloudy challenges and opportunities of cloud computing, International Journal of Communication Systems 25 (2012), 796–819.

Rajavel and

Thangarathanam , SLAOCMS: A layered architecture of SLA oriented cloud management system for achieving agreement during resource failure, Advances in Intelligent Systems and Computing 236 (2014), 801–809.

10.

Buyya ,

S.K.

Garg and

R.N.

Calheiros , SLA-Oriented Resource Provisioning for Cloud Computing: Challenges, Architecture, and Solutions, IEEE International Conference on Cloud and Service Computing, 2011, pp. 1–10.

11.

Corradi and

O.F.

Rana , Editorial: The management of cloud systems, Future Generation Computer Systems 32(2014),24–26.

12.

A.V.

Dastjerdi and

Buyya , An autonomous time-dependent SLA negotiation strategy for cloud computing, The Computer Journal 58 (2015), 1–15.

13.

Venticinque ,

Tasquier and

B.D.

Martino , Agents based Cloud Computing Interface for Resource Provisioning and Management, International Conference on Complex, Intelligent, and Software Intensive Systems, 2012, pp. 249–256.

14.

K.M.

Sim , Towards complex negotiation for cloud economy, International Conference on Advances in Grid and Pervasive Computing, LNCS 6104, 2010, pp. 395–406.

15.

Adabi ,

Movaghar ,

A.M.

Rahmani and

Beigy , Market_based grid resource allocation using new negotiation model, Journal of Network and Computer Applications 36 (2013), 543–565.

16.

Brandic ,

Music and

Dustdar , Service mediation and negotiation bootstrapping as first achievements towards self-adaptable grid and cloud services, Grids and Service-Oriented Architectures for Service Level Agreements, 2010, pp. 119–132.

17.

Chieng ,

Marshall and

Parr , SLA brokering and bandwidth reservation negotiation schemes for QoS-aware internet, IEEE Transactions on Network and Service Management 2 (2005), 39–49.

18.

Groleat and

Pouyllau , Distributed learning algorithms for inter-NSP SLA negotiation management, IEEE Transactions on Network and Service Management 9 (2012), 433–445.

19.

Liu ,

Zhang ,

Yang and

Chena , CCBKE-Session key negotiation for fast and secure scheduling of scientific applications in cloud computing, Future Generation Computer Systems 29 (2013), 1300–1308.

20.

K.M.

Sim , Complex and concurrent negotiations for multi- [36] ple interrelated e-markets, IEEE Transaction on Cybernetics 43 (2013), 230–245.

21.

Rahwan ,

S.D.

Ramchurn ,

N.R.

Jennings ,

Mcburney ,

Parsons and

Sonenberg , Arumentation-based nego-tiation, The Knowledge Engineering Review 18 (2003), 343–375.

22.

Zheng ,

Martin and

Brohman , Cloud Service Negotiation: Concession vs. Tradeoff Approaches, 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing, 2012, pp. 515–522.

23.

Yan ,

Kowalczyk ,

Lin ,

M.B.

Chhetri ,

S.K.

Goh and

Zhang , Autonomous service level agreement negotiation for service composition provision, Future Generation Computer Systems 23 (2007), 748–759.

24.

G.C.

Silaghi ,

L.D.

Serban and

C.M.

Litan ,A time-constrained SLA negotiation strategy in competitive computational grids, Future Generation Computer Systems 28 (2012), 1303–1315.

25.

Adabi ,

Movaghar ,

A.M.

Rahmani and

Beigy ,Negotiation strategies considering market, time and behavior functions for resource allocation in computational grid,Journal of Supercomputing 66 (2012), 1–40.

26.

Gwak and

K.M.

Sim , An augmented EDA with dynamic diversity control and local neighborhood search for coevolution of optimal negotiation strategies, Applied Intelligence 38 (2013), 600–619.

27.

Pooyandeh and

D.J.

Marceau , Incorporating Bayesian learning in agent-based simulation of stakeholders’ negotiation, Computers, Environment and Urban Systems 48(2014),73–85.

28.

Son and

K.M.

Sim , A negotiation mechanism that facilitates the price-timeslot QoS negotiation for establishing SLAs of cloud service reservation, Communicati in Computer and Information Science 136 (2011 432–446.

29.

Gwak and

K.M.

Sim , A novel method for coevolving PS-optimizing negotiation strategies using improved diversity controlling EDAs, Applied Intelligence 38 (2013), 384–417.

30.

Son and

K.M.

Sim , A price-and-time-slot-negotiation mechanism for cloud service reservations, IEEE Transaction on Systems, Man, and Cybernetics - Part B: Cybernetics 42 (2012), 713–728.

31.

Son and

K.M.

Sim , Adaptive and similarity-based tradeoff algorithms in a price-timeslot-QoS negotiation system to establish cloud SLAs, Information Systems Frontiers 17(2015),565–589.

32.

Zeng and

Sycara , Bayesian learning in negotiation, International Journal of Human-Computer Studies 48 (1998), 125–141.

33.

D.C.

Moosmayer ,

A.Y.

Chong ,

M.J.

Liu and

Schuppar ,A neural network approach to predicting price negotiation outcomes in business-to-business contexts,Expert Systemswith Applications 40 (2013), 3028–3035.

34.

Wang ,

T.N.

Wong and

Yu , A computational model for multi-agent Ecommerce negotiations with adaptive negotiation behaviors, Journal of Computational Science 4 (2013), 135–143.

35.

Rajavel and

Thangarathanam , ADSLANF: A negotiation framework for the cloud management system using Bulk Negotiation Behavioural Learning approach, Turkish Journal of Electrical Engineering & Computer Sciences 25 (2017), 1–50.

36.

Rajavel and

Thangarathanam , Adaptive probabilistic behavioural learning system for the effective behavioural decision in cloud trading negotiation market, Future Generation Computer Systems 58 (2016), 29–41.

37.

Pouyllau and

Carofiglio , Inter-carrier SLA negotiation using Q-learning, Telecommunication Systems 52 (2013), 611–622.

38.

Rapoport ,

T.E.

Daniel and

D.A.

Seale , Reinforcement-based adaptive learning in asymmetric two-person bargaining with incomplete information, Experimental Economics 1 (1998), 221–253.

39.

Monteserin and

Amandi , A reinforcement learning approach to improve the argument selection effectiveness in argumentation-based negotiation, Expert Systems with Applications 40 (2013), 2182–2188.

40.

Gerding ,

D.V.

Bragt and

H.L.

Poutre , Multi-issue negotiation processes by evolutionary simulation, validation and social extensions, Computational Economics 22 (2003), 39–63.

41.

Gwak ,

K.M.

Sim and

Jeon , Novel dynamic diversity controlling EAs for coevolving optimal negotiation strategies, Information Sciences 273 (2014), 1–32.

42.

Hsu ,

Kao ,

V.L.

Ho ,

Li and

K.R.

Lai , An agent-based fuzzy constraint-directed negotiation model for solving supply chain planning and scheduling problems, Applied Soft Computting 48 (2016), 703–715.

43.

Kolomvatsos ,

Panagidi ,

Neokosmidis ,

outas and

Hadjiefthymiades , Automated concurrent negotiations: An artificial bee colony approach, Electronic Commerce Research and Applications 19 (2016), 56–69.

44.

Li ,

C.S.

Yeo ,

Hsu ,

Yu and

K.R.

Lai , Agent-based fuzzy constraint-directed negotiation for service level agreements in cloud computing, Cluster Computing 21 (2018), 1349–1363.

45.

Shojaiemehr and

M.K.

Rafsanjani , A supplier offer modification approach based on fuzzy systems for automated negotiation in e-commerce, Information Systems Frontiers 20 (2018),143–160.

46.

K.M.

Sim , Evolving fuzzy rules for relaxed-criteria negotiation, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) 38 (2008),1486–1500.

47.

C. W.

Khuen ,

C.H.

Yong and

Haron , A framework for multi-agent negotiation system using adaptive fuzzy logic in resource allocation, International Journal of Information Technology 11 (2005), 35–49.

48.

R.G.

Smith , The contract net protocol: High-level communication and control in a distributed problem solver, IEEE Transactions on Computers C-29 (1980), 1104–1113.

49.

Venugopal ,

Chu and

Buyya , A negotiation mechanism for advance resource reservations using the alternate offers protocol, International Workshop on Quality of Service, Enschede (2008), 40–49.

50.

Czajkowski ,

Foster and

Kesselman , SNAP: A protocol for negotiating service level agreements and coordinating resource management in distributed systems, Job Scheduling Strategies for Parallel Processing 2537 (2002), 153–183.

51.

Klenk ,

Beck-Greinwald ,

Angst and

Carle , Iterative multi-party agreement negotiation for establishing collaborations, Service Oriented Computing and Applications 6 (2012), 321–335.

52.

Lang and

Fink , Negotiating in dynamic environments: Time efficient automated negotiations by means of combinatorial auctions, Evolving Systems 3 (2012), 189–201.

53.

K.M.

Sim , Agent-based cloud computing, IEEE Transactions on Service Computing 5 (2012), 564–577.

54.

Russell ,

M.L.

Littman and

Trappe , Integrating machine learning in ad hoc routing: A wireless adaptive routing protocol, International Journal of Communication Systems 24 (2011), 950–966.

55.

Shojaiemehr ,

A.M.

Rahmani and

N.N.

Qader , Cloud computing service negotiation: A systematic review, Computer Standards & Interfaces 55 (2018), 196–206.

56.

Sanchez-Anguix ,

Julian ,

Botti and

García-Fornes , Tasks for agent-based negotiation teams: Analysis, review, and challenges, Engineering Applications of Artificial Intelligence 26 (2013), 2480–2494.

57.

Wooldridge , An Introduction to Multiagent Systems. (2nd ed.). John Wiley & Sons, United Kingdom, Chapter 2, 2009.