Академический Документы
Профессиональный Документы
Культура Документы
277-285
Manuscript received December 20 , 2016; revised June, 2017. DOI: 10.1515/eletel-2017-0037
Abstract—Fairness provisioning in heterogeneous networks is a prime attribute scheduling algorithm approach is proposed, where
issue for high-rate data flow, wherein the inter-connectivity property among these parameters are used for resource scheduling. This multi-
different communication devices provides higher throughput. In Hetnet, attribute approach is significant because it controls the allocated
optimal resource utilization is required for efficient resource usage. Proper
resource allocation in such a network led to higher data flow performance data rate based on an optimized utility function in the HetNet.
for real-time applications. In view of optimal resource allocation, a resource This optimization is proved wrt. optimal rate allocation over
utilization approach for a reconfigurable cognitive device with spectrum maximization of the spectrum efficiency and fairness factor. In
sensing capability is proposed in this paper. The allocation of the data flow order to be fair to the user, an interference alignment (IA)
rate at device level is proposed for optimization of network fairness in a
heterogeneous network. A dynamic approach of rate-inference optimization
scheme for resource allocation was presented in [2]. A spectral
is proposed to provide fairness in dynamic data traffic conditions. The coexistence mechanism to mitigate the interference in
simulation results validate the improvement in offered quality in comparison heterogeneous networks was presented. The resource allocation
to multi-attribute optimization. in terms of optimal interference management was advised. In
[3], the optimality issue of resource allocation with respect to
Keywords—rate-inference optimization, heterogeneous network, flow traffic density was outlined. The offered capacity for
control, spectrum utilization
transmission was governed by the solution of a continuous
linear programming (LP) approach, where a solution is derived
I. INTRODUCTION by a scheduler unit based on the number of mobiles in the
QoS in heterogeneous networks, a quality objective with regard for the developed approach. The conventional model of the
to an interference constraint across the cell coverage is multi-attribute optimization approach, based on spectral
presented in [13] . The QoS parameters are defined as a function efficiency and fairness as outlined in [1], is presented in section
of link bandwidth, node buffer and queuing service and are III. Section IV outlines the proposed spectrum sense
employed at the router level in a heterogeneous network. In optimization approach and rate flow control for heterogeneous
[14], an end-to-end QoS framework for heterogeneous networks. Section V illustrates the experimental results derived
networks is focused on. The objective of QoS provisioning was from the proposed approach in comparison to conventional
made over multiple autonomous systems. In [15],16], a routing modelling.
algorithm based on proactive coding was suggested, and a QoS-
driven control selection was outlined. A general backward II. SYSTEM MODEL
utility formation was introduced to observe the dynamic QoS For the communication of data, a cluster of subnetworks
variation in the network. With respect to the resource variation capable of wireless transmission is used. An instance of the
issue and governing of QoS, multi-hop communication with network model following the heterogeneous network
regard to homogeneous and heterogeneous networks was representation is shown in Fig. 1.
presented in [17]. The variant parameter of a micro-site with
regard to a traffic load condition was analysed. In [18],19], IP-
oriented QoS signalling as a function of end-to-end QoS
offering and seamless mobility for joint resource management
was presented. The approach of node automaticity and quality
monitoring was presented with mobility concerns for QoS
offering. The method derived from supporting various services
in different architectures of heterogeneous networks. In
providing QoS, spectrum allocation based on the device level is
presented in [20]. An energy-efficient cooperative spectrum
sensing strategy for immune noise uncertainty was presented.
Coarse and fine sensing schemes are employed to achieve Fig. 1. Heterogeneous Architecture
energy efficiency in HetNets. A relation property of noise
immunity over power consumption was derived. To offer The network is defined with a set of communication
resource utilization via sharing, a device-to-device-level coding terminals, defined as cognitive user units (CuEs), where each
is presented in [21] based on a spectrum partitioning approach, device has the property of cognitive operation. In this network,
and an improved spectrum efficiency for D2D and cellular the communication terminals are linked with access points
networks was derived. Under quality-of-service governance, (APi), where each of these access points is set with different
the spectrum is divided into several subfrequencies, and each communication standards. The access network is defined with
band is allocated a D2D pair for transmission. Optimality was a Wi-Fi terminal, a cellular terminal and a WiMAX link point.
observed for maximization of D2D transmission power. The In the process of communication, a communication terminal
approach was developed under the constraint of offering QoS requests access. Based on the available access terminal, an
to the cellular transmission conformation. In [22], a device- access link is selected to offer maximum resources. The
level coding using cognitive radio units (CRNs) is presented. selected access terminal is then linked with the communication
An active resource allocation problem for heterogeneous terminal via a GRCU (global resource control unit). To perform
networks under imperfect channel conditions was presented the resource allocation, the communication terminal registers
The optimality issue was resolved by applying an optimal the selected access terminal to the GRCU, and then a handshake
power allocation problem under a known channel condition. In process is carried out between the GRCU and the
the suggested approach under network abstraction, the node communication terminal, confirming the allocation. Once the
operating as a router was controlled via different approaches to allocation is confirmed, communication is established from the
achieve the objective of higher performance in a heterogeneous communication terminal to the destination terminal via the
network. In [23], a bandwidth buffering mechanism addressing selected access point, controlled by the GRCU. The flow
the allocation issue and different queuing services to provide control of the heterogeneous network is illustrated in Fig. 2.
end-to-end QoS provision in heterogeneous networks was
suggested. isA self-clocked fair queuing mechanism for
providing QoS governance was presented, and the issues of the
end-to-end requirement and user connectivity were addressed.
While the provision of optimal quality of service and fairness
was solved in different aspects, multi-objective control for
HetNets is still to be developed to achieve optimal equilibrium
in resource utilization with offered QoS and network operation.
In this paper, a new approach for quality-governed data flow
operation is outlined to achieve efficient operational
performance in heterogeneous networks. An integration
approach for device performance and network control is
proposed to reach the target of higher performance in a
heterogeneous network. The remainder of this paper is
organized as follows. Section II outlines the system model used Fig. 2. Control flow diagram of the considered Heterogeneous Network
OPTIMAL SPECTRUM UTILIZATION AND FLOW CONTROLLING IN HETEROGENEOUS NETWORK WITH RECONFIGURABLE DEVICES 279
During the process of communication, a terminal access was where 𝜔𝑢𝑡 is the utility function for the battery life of each
granted by the GRCU, registering the sensed access point with user. These multiple attributes are controlled, to obtain an
the highest available resource. The resource allocations take efficient coding for spectral efficiency, fairness and long life of
place in the GRCU unit, and an optimal data rate is allocated nodes in a heterogeneous network. However, the concerns of
based on the current state of the registered access units. For device efficiency and service quality have not been focused on.
resource allocation, the GRCU unit operates with a multi- The optimization is derived based on the utility factors of these
attribute optimization approach as outlined in [1]. The attributes, and the control is made wrt. the allocated rate at time
conventional approach of multi-attribute optimization is t. To achieve higher efficiency in resource utilization with a
outlined in the next section. quality concern in such Heterogeneous Networks, a quality-
governed approach with device reconfiguration is proposed.
III. MULTI-ATTRIBUTE OPTIMIZATION APPROACH [1]
To provide efficient coding for resource allocation in IV. PROPOSED OPTIMIZATION ATTRIBUTE
heterogeneous networks considering Wi-Fi, 4G
(LTE/WiMAX), and 3G (HSPA/EVDO) architectures, a multi- A. Spectrum sense optimization
attribute optimization approach is suggested in [1]. In the In the present approach, during the initialization process, the
process of the multi-attribute optimization approach, 3 factors cognitive user equipment (CuE) performs a search operation in
were considered: 1) Spectral efficiency, 2) Fairness and 3) Connectivity Access Networks (CANs) based on the order of
Battery life. Spectral efficiency is calculated as the utility network preference. In this approach, the order of preference is
function of the allocated rate over the total spectrum 𝑘 available set as Wi-Fi, WiMAX and Cellular networks. For each of these
at time t for u users belonging to a set of users U, i.e., networks, a utility factor is calculated, and the network offering
∑𝑢∈𝑈 𝑟𝑢𝑡 the higher fairness factor is chosen. The selected CAN is
𝛾𝑡 = (1)
𝑘 registered with the GRC, and communication is then formed
where the allocated rate 𝑟𝑢𝑡 is defined by over the registered CAN. In the process of communication, the
GRC allocates the spectrum for communication under the
𝑟𝑢𝑡 𝑡
= ∑𝑎∈𝐴 𝑥𝑢𝑎 𝑡
∗ 𝑟𝑢𝑎 (2)
limitation of 𝛾𝑢𝑡𝑖𝑙 ∈ [0,1]. The rate allocation is based on the
where A is the set of BS/APs registered and 𝑎 ∈ 𝐴 is the radio demanded spectrum to the GRC, where the GRC performs
offered to A at time t. multi-attribute resource allocation to allocate resources to the
The optimal spectral allocation is obtained by optimizing the requested unit. However, with the communication process, the
allocation rate, defined by the maximum sum rate (MSR) spectrum is dynamically engaged and released based on the
optimization problem. However, the MSR has a limitation for
demanded service. The vacant spectrum can be utilised for
the global user resource allocation problem, which yields an
current active users towards extra spectrum allocation to
unfairness in spectrum allocation. To optimize the spectrum
𝑡 improve service quality. As the devices are cognitive in nature,
allocation, a utility factor 𝛾𝑢𝑡𝑖𝑙 is used, which is defined by [1]
the cognitive property of resource allocation can be used for
as
𝑡
𝛾𝑡 −𝛾𝑚𝑖𝑛
these spectrum allocations. The initial spectrum allocation for
𝑡
𝛾𝑢𝑡𝑖𝑙 = 𝑡 𝑡 (3) user u € U is given by the utility value 𝛾𝑈𝑡 . On the release of a
𝛾𝑚𝑎𝑥 −𝛾𝑚𝑖𝑛
𝑡
primary allocation rate 𝑟𝑖𝑡 , the available spectrum is defined by
where 𝛾𝑚𝑎𝑥 is the maximum achievable spectral efficiency
utility function and 𝛾𝑚𝑖𝑛 𝑡
is the minimum spectral utility
𝑡
𝛾𝑎𝑣1 = ∑𝑢∈𝑈 𝛾𝑢𝑡 − 𝑟𝑖𝑡 (7)
𝑡
function. When the network achieves 𝛾𝑚𝑎𝑥 efficiency, the This available spectrum is called the secondary spectrum and
utility function is given a value of 1, while the value is 0 when can be utilised by the active primary users (𝑝𝑢 ). On the request
𝑡
the achieved spectral efficiency is 𝛾𝑚𝑖𝑛 . To provide a fair for this secondary spectrum, the GRCU computes the spectral
platform, the allocated rate is controlled by a fairness utility interference energy, 𝐸, for n active user as defined by [20] as
𝑡 𝑡
function 𝜃𝑢𝑡𝑖𝑙 . This utility function 𝜃𝑢𝑡𝑖𝑙 is defined by [1] as, 𝑦
𝐸 = ∑𝑛𝑖=1( 𝑖)𝑃 (8)
|𝐵𝑈 𝑡 | 𝜎
𝑡
𝜃𝑢𝑡𝑖𝑙 = 1− (4)
|𝑈| where 𝑦𝑖 is the received information and σ is the observed
The fairness utility function is defined as a function of user channel variance. Parameter ‘P’ is varied over a range of 1-10
blockage, where it takes a value of 1 when the number of to derive the energy variation wrt. the interference variation in
blocked users (BUs) = 0 and takes a value of 0 when the total the channel. To allocate the available secondary spectrum, the
blockage is maximal. For long-term fairness, the utility function obtained energy ‘E’ is used as the allocation parameter. In the
is defined by [1] as data flow process, the optimal encoding mode can be
(∑ ∑𝑡 𝑟𝑢𝑡 )2 determined by reaching the best trade-off between the amount
𝜙𝑢𝑡𝑖𝑙 = |𝑈|∗∑𝑢∈𝑈 𝑡 2 (5)
𝑢∈𝑈(∑𝑡 𝑟𝑢 ) of coding bits and the obtained service quality. This problem
where the long-term fairness is derived by the aggregated can be modelled as a maximum sum rate optimization problem
observation of long-term small rate allocations. As a third [1]. This MSR optimization is developed to achieve the
attribute, battery life is considered for optimal resource objective of fairness of resource utilization in a heterogeneous
allocation. The utility factor under this objective is defined by network. The MSR optimization is governed for allocation of
[1] as the maximum rate of data flow achieving maximum spectral
𝑡 −𝜔𝑡
∑𝑢∈𝑈 𝜔𝑢
𝑡
𝜔𝑢𝑡𝑖𝑙 =1− 𝑡
𝑚𝑖𝑛
(6) efficiency. To achieve better service quality for the spectrum
𝑡
𝜔𝑚𝑎𝑥 −𝜔𝑚𝑖𝑛
allocation, a distribution optimization is proposed. To govern
280 S. S. HASHMI, S. A. SATTAR, K. SOUNDARARAJAN
quality, a cost function with distortion and rate allocation is this objective, a rate interference approach is proposed in
defined, subjected to minimization of distortion (D), subjected consideration of the data flow per node for the allocated
to an allocation free secondary spectrum ‘R’ constraint to the spectrum. This update in rate allocation results in higher
𝑡
allocation rate 𝑅𝑐 , which is given by 𝛾𝑎𝑣1 . The optimization spectrum efficiency in consideration of the cognitive property
operation of the spectrum is defined as of the CuE. In addition to spectrum efficiency, a fairness
min{𝐷} 𝑠𝑢𝑏𝑗𝑒𝑐𝑡 𝑡𝑜 𝑅 ≤ 𝑅𝑐 (9) optimization is applied to achieve a higher service level in
{𝑚} Heterogeneous Networks.
which indicates that the encoder should satisfy the minimum B. Rate-inference optimization
distortion ’D.’
The allocation of spectrum to a user with the data rate While the suggested approach of rate allocation for the
satisfying the instantaneous fairness utility function θt 𝑢𝑡𝑖𝑙 was fairness approach is observed to be very effective in data flow,
developed [1]. However, in real-time data traffic, such as video when applied to the data flow over a wireless channel with
streaming or VoIP services, the fairness is defined by the multiple sources, it is observed that the rate of transmission and
offered BER, controlled by the allocation of data rate. In the offered quality need to be optimized. In the approach of data
practical usage, in the network abstraction layer (NAL), the traffic data flow, due to remote capturing, the allocated
network operates at a different level to provide proper resources of power and spectrum will be limited. In such a case,
allocation of resources under different data traffic conditions. the proper utilization of resources and intermediate node
In wireless communication, the transmission channel is time- support is highly required to achieve greater performance. The
varying and distortive in nature. For minimization of error approach of observing for error allocation coding is simpler,and
during signal propagation, an independent channel model is effective for service data flow, but without control of the flow,
used, deriving the bit error rate (BER) of the transmission this allocation may not result in effective service quality at
channel. The packet loss probability ‘𝜌’ during the transmission receiving units due to the incurred latency issue. Hence, the
of ‘L’ bits for a given BER is defined by access control is also needed with spectrum allocation; with this
objective, a buffer-based management scheme to control the
𝜌 = 1 − (1 − 𝑏𝑒𝑟)𝐿 (10) rate of allocation was proposed. As in the proposed work, each
𝑡
The quality utility function 𝛾𝑖𝑛𝑡 is then defined by intermediate node is considered to be a router, and each node
𝑡 −𝛾 𝑡
𝑡 𝛾𝜑 𝑚𝑖𝑛 encounters multiple data traffic at this node and observes heavy
𝛾𝑢𝑡𝑖𝑙𝜑
= 𝑡 𝑡 (11)
𝛾𝑚𝑎𝑥 −𝛾𝑚𝑖𝑛 congestion. To overcome this issue, buffer management is
𝑡
∑𝑢∈𝑈 𝛾𝑢𝜑 applied. The cross-layer optimization of service stream data
where, 𝛾𝜑𝑡 = (12) traffic at the router level was proposed in [14]. The approach of
𝑘
𝑡
𝛾𝑢𝜑 𝑡
= (𝛾𝑢𝑎 + 𝐿) (13) coding was introduced at the network abstraction layer (NAL),
𝑡 𝑡
where the buffer-based congestion control follows buffer
where 𝛾𝑢𝜑 is the allocated rate with the primary data rate 𝛾𝑢𝑎 management and the relative quality of service (QoS) is
and secondary spectrum of L bits. During the transmission mapped to schedule the rate of data traffic flow. The cross-layer
process, the packet stream is divided into L-bit slices approach processes the measured buffer length and derives the
represented as 𝑠𝑛,𝑚 . For the Tx of the mth slice in the nth frame, packet En buffer or doping probability based on receiving data
the BER is defined by 𝑏𝑒𝑟𝑛,𝑚 , which is the channel BER for the traffic. At the Service coding layer, the communication data are
transmission, and 𝜌𝑛,𝑚 is the packet loss rate for a slice 𝑠𝑛,𝑚 . blocked into slices and passed to NAL for rate allocation. The
To find the optimal bits of length ‘L,’ minimizing the packet method computes the utility function ‘γt 𝑢𝑡𝑖𝑙 ’ as
loss 𝜌, the optimization was subjected to min (𝜌) and max(L).
𝑡
min(𝑙) ⇒ max(𝐿) (14) 0; 𝜃𝑢𝑡𝑖𝑙 < γt 𝑚𝑖𝑛
𝑡
To generate the data transmitting bits ‘L’ for demand service 1; 𝜃𝑢𝑡𝑖𝑙 > γt 𝑚𝑎𝑥
γt 𝑢𝑡𝑖𝑙 = (15)
quality, given by required BER. An optimal value of L is γt 𝑢𝑡𝑖𝑙 −γt 𝑚𝑖𝑛
derived, obtaining minimal packet loss. In the operation of rate 𝑚𝑎𝑥𝑝 × ; 𝑜𝑡ℎ𝑒𝑟𝑤𝑖𝑠𝑒
γt 𝑚𝑎𝑥 −γt 𝑚𝑖𝑛
{
allocation, the fairness utility function is governed by two levels
of maximum or minimum spectral utility functions. The utility where 𝑚𝑎𝑥𝑝 is defined as the maximum number of packets
component in this case is normalized to two levels, [0, 1]. While exchanged over a node. In the approach of cross-layer
𝑡
this is developed with the concern of maximum or minimum modelling, 𝜃𝑢𝑡𝑖𝑙 is updated by 1 − 𝜙 𝑡 , where 𝜙 𝑡 is the update
spectrum utilization, the traffic flow per node is not considered. applied in period t and 𝜙 is the constant value 1.001 defined as
Data traffic on a heterogeneous node is dynamic in nature, and the random exponential marking.
the two-level normalization process is only for spectrum The update factor is occasionally updated according to the
utilization, wherein this is limited at the device level. The issue average buffer length and the input rate and output rate of the
arises when the nodes are highly congested due to variable data buffer. This approach controls the flow of data traffic by
traffic from different sources, flowing in at different data rates. accepting or dropping service packets based on the probability
Though the optimal spectrum utilization is used, if the nodes index 𝜙 𝑡 and the importance of the packet. The factor is
are blocked at the node level, the network will fail to deliver the incremented if the input rate exceeds the output rate and
demanded service rate. Hence, in addition to spectrum utility decremented otherwise. The approach of traffic flow control at
function, a device-level optimization is required. To achieve node level is defined by the following algorithm.
OPTIMAL SPECTRUM UTILIZATION AND FLOW CONTROLLING IN HETEROGENEOUS NETWORK WITH RECONFIGURABLE DEVICES 281
where
𝑅𝑎𝑙𝑙𝑜𝑐 (𝑡)= allocated data rate
Nodes in Random Topology
𝑅(𝑡) = full rate 120 12
∆𝑡 = incremental data rate 8
15
22
γt 𝑚𝑖𝑛 = minimum limit of the buffer 100 30
24
γt 𝑚𝑎𝑥 = maximum limit of the buffer 28
25
2
21
80 20
27
1
17
In the above equation, the allocated data rate is varying with 60 7
18
respect to the node congestion level. If the current buffer length 6
4
3
is at the minimum level, the data will be allocated with an 29
40 9
increment of ‘∆𝑡.’ If the current buffer length in a node is in 13
19
between the minimum and maximum utility level, the allocated 16
10
20 11
data rate is governed by the traffic flow. Similarly, if the current 23
26
buffer length exceeds the maximum limit, which represents a 14
5
0
traffic flow, the allocated data rate will vary according to the 0 20 40 60 80 100 120
𝑡
offered traffic utility function 𝜃𝑢𝑡𝑖𝑙 . Fig. 3. Randomly distributed access points for a terminal density of 30
V. EXPERIMENTAL RESULTS
Network with range Circle
To evaluate the operational performance of the proposed 120 12
8
approach, a MATLAB simulation model is developed, where a 15
22
24
distributed wireless Heterogeneous network is simulated with 100 30
28
the proposed optimization attribute of the rate flow control 25
2
approach and compared with the conventional approach of 21
80 20
27
multi-attribute optimization [1] and QoS-based buffer 1
17
management [14]. For the simulation, a network is defined in 18
60 7
Table I with the following network parameters: 6
4
3
For the analysis of the proposed rate flow attribute-based 40
29
9
coding approach over the conventional multi-attribute control 13
19
approach, different networks with varying node distributions 16
10
20 11
are used. Node densities of 30 and 40 terminals are simulated, 23
26
and the obtained observations are illustrated below. 14
5
0
With the network under consideration, a network with 0 20 40 60 80 100 120
randomly distributed terminals is modelled. The randomly Fig. 4. Network with coverage regions marked
distributed network architecture used for the simulation is
282 S. S. HASHMI, S. A. SATTAR, K. SOUNDARARAJAN
Overhead
18
60 7 25
6
4
3 20
29
40 9
13 15
19
16
10 10
20 11
26
23 5
14
5
0 0
0 20 40 60 80 100 120 1 1.5 2 2.5 3 3.5 4 4.5 5
iteration
Fig. 5. Network with 1-Hop neighbours
Fig. 7. Network overhead with packet forwarding
Selected route
120 12 Network Throughput Plot
8 100
15
22 proposed Optimization scheme
24 90 Multi attribute optimization [1]
100 30
28 QoS buffer control scheme[14]
25 80
2
21
80 20 70
27
1 60
17
Throughput
18
60 7 50
6
4
3 40
29
40 9
13 30
19
16
10 20
20 11
26
23 10
14
5
0 0
0 20 40 60 80 100 120 1 1.5 2 2.5 3 3.5 4 4.5 5
iteration
Fig. 6. Network with a selected path for communication Fig. 8. Network throughput with the forwarding of packets
The network with 1-hop neighbours is illustrated in Fig. 5. Due to the rate flow attribute coding, the congestion is
Each node computes the distance of nodes with all other controlled over a limit of buffering in contrast to buffering to a
terminals, and the distance below the communication range is higher level of buffer queue. Due to early control and rate flow
defined as a 1-hop neighbour for the refereeing node. In the attribute estimation, the overhead for such a network is
process of neighbour discovery, the Terminal distances falling observed to be minimized with the forwarding of packets.
below the communication range are declared as direct The network throughput for the developed network is
neighbours, and each node exchanges their one-hop neighbour illustrated in Fig. 8. Due to the higher data rate of packet
details to formulate a link network. forwarding in the proposed approach, it is observed that more
With the one-hop neighbour list, an AODV routing protocol packets are received, hence resulting in higher throughout in the
is developed to evaluate the route from a given source to the network. The throughput of these simulated systems is defined
destination. The routes are generated via the data exchange as the number of packets generated over packets received for an
from source to destination in a broadcasting manner. A forward observed communication time period. The throughput is
and backward tracing method is applied to obtain the route observed to be the same for the first 1-hop forwarding, and in
parameters used for forwarding packets. Among the generated such a case, the buffering is observed to be minimized; hence,
routes, a path with minimum hops (i.e., the shortest path) is the allocated data rate is improved. However, as the amount of
selected for communication. The selected path for the packet forwarding increases, the congestion level increases,
developed network is as illustrated in Fig. 6. resulting in the decrement of throughput, which is maintained
In the process of communication, data are forwarded along to be higher and linear in the case of rate flow attribute coding.
the path selected from source to destination via intermediate The observed end-to-end delay factor for the developed
terminals. During the process of data exchange, the packets are system is presented in Fig. 9. The delay for the rate control
buffered, and based on the computed congestion level, packets approach is higher than that of the rate flow attribute coding, as
are forwarded. To transfer the packets, the per node forwarding the buffering of data at each node is minimized at the node level
data rate is computed based on the congestion level, and and the packets are released faster. The congestion level in such
allocated data rate packets are released or queued in the node. coding is developed in a rate flow attribute manner, wherein
The queued packets internally build an overhead in the network. packets are buffered in the queue based on the rate control
This overhead is defined as the number of packets queued for approach, which builds the forwarding delay for each node.
processing at each node. With the forwarding of packets from The buffered Q length for such a system is presented in Fig.
source to destination at each intermediate node, packets are 10. The Q lengths are measured as the volumes of data packets
buffered; the overhead observed is hence presented in Fig. 7. buffered with an increase in forwarding of data packets. It is
OPTIMAL SPECTRUM UTILIZATION AND FLOW CONTROLLING IN HETEROGENEOUS NETWORK WITH RECONFIGURABLE DEVICES 283
observed that the Q length of buffering is reduced for rate flow Allocation Rate Plot
25
attribute coding due to the increment in the data rate. The proposed Optimization scheme
queuing is, however, observed to be equal in the initial Multi attribute optimization [1]
QoS buffer control scheme[14]
communication phase and gradually increases with the 20
forwarding of packets.
11 2937
15
80
28 6
3 39
18 38
10 60 13 24
31 23
40 4 2
5 40 15
20 27
20 7 25
19 33
0
0
1
1 1.5 2 2.5 3 3.5 4 4.5 5
0 20 40 60 80 100 120 140 160
iteration
Fig. 9. Observed delay for the simulated network with a node density of 30 Fig. 12. Network topology for randomly distributed terminals with a node
density of 40
Queue Length Plot
100
proposed Optimization scheme
90 Multi attribute optimization [1]
QoS buffer control scheme[14]
80
70
Queue Length
60
50
40
30
20
10
0
1 1.5 2 2.5 3 3.5 4 4.5 5
iteration
Fig. 10. Buffered Q_length with an increase in forwarded packets Fig. 13. Marked coverage regions for each node in the network
284 S. S. HASHMI, S. A. SATTAR, K. SOUNDARARAJAN
Throughput
11 2937
50
80
28 6
3 39 40
18 38
60 13 24 30
31 23
40 4 2 20
40 15
20 27 10
20 7 25
19 33 0
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
0
1 iteration
0 20 40 60 80 100 120 140 160
Fig. 17. Network throughput for the developed network for a node density
Fig. 14. Network with 1-hop neighbour links of 40
40 4 2
40 15 15
20 27
20 7 25 10
19 33
0
1
0 20 40 60 80 100 120 140 160 5
The node distribution, coverage range, neighbour links and Fig. 18. End-to-end delay for the developed network
selected path for communication are as illustrated in Figures
Queue Length Plot
12-15, respectively. 100
proposed Optimization scheme
90 Multi attribute optimization [1]
QoS buffer control scheme[14]
Network Overhead Plot 80
80
proposed Optimization scheme 70
70 Multi attribute optimization [1]
Queue Length
60 50
40
50
Overhead
30
40
20
30 10
0
20 1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
iteration
10
Fig. 19. Observed Q-Length at the terminal for the developed network
Allocation Rate Plot
0 25
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
iteration proposed Optimization scheme
Multi attribute optimization [1]
QoS buffer control scheme[14]
20
Fig. 16. Network overhead with an increase in the forwarding of packets
Allocated Data Rate (B/s)
15
The network overhead in such a case is observed to be
reduced from 122 to 71. The overhead is minimized due to the
10
existence of more terminals to exchange the data, which results
in faster data exchange. However, due to the non-optimal rate
5
control approach, the overhead is observed to be higher in the
rate control approach in comparison to the rate flow attribute
0
approach. 1 1.5 2 2.5 3 3.5
iteration
4 4.5 5 5.5 6
The observed delay factor is minimized from 10 to 7 for [5] A. Khandekar, N. Bhushan, J. Tingfang, V. Vanghi, “LTE-advanced:
Heterogeneous networks”, In Wireless Conference (EW), 2010
higher node density, the Q_length is decreased from 42 to 25,
European (pp. 978-982), IEEE, 2010.
and the allocable data rate is increased from 5 to 7 for the [6] W. Mansouri, K. Mnif, F. Zarai, M. S. Obaidat, “A new multi-rat
developed rate flow attribute method. The obtained scheduling algorithm for Heterogeneous wireless networks”, Journal of
observations for the delay, Q_length and allocated data rate is Systems and Software, Elsevier, 2015.
[7] F. W. Karam, T. Jensen, "On schemes for supporting QoS and mobility
presented in Figures 18-20. The observation made for a node
in heterogeneous networks." Proceedings of the 7th International
density of 40 terminals is presented in Table II. The allocated Conference on Frontiers of Information Technology. ACM, 2009.
data rate could be observed to be improvised in rate flow [8] M. Liu, Y. Teng, M. Song. "Tradeoff Analysis between Spectrum
attribute-based coding as presented in Table II. The average Efficiency and Energy Efficiency in Heterogeneous Networks (HetNets)
Using Bias Factor." Journal of Communications 10.10, 2015.
queue length is hence reduced from 29 to 19 in this case. The
[9] K. Son, E. Oh, B. Krishnamachari. "Energy-efficient design of
network throughput is observed to be increased from 13.34 to heterogeneous cellular networks from deployment to
53.36 in such case. The delay is minimized from 13 to 7 for the operation."Computer Networks vol. 78, pp. 95-106, 2015.
rate flow attribute-based control approach in comparison to the [10] G. He, S. Zhang,, Y. Chen, S. Xu, “Spectrum efficiency and energy
efficiency tradeoff for heterogeneous wireless networks”, In 2013 IEEE
rate control approach.
Wireless Communications and Networking Conference (WCNC), pp.
TABLE II
2570-2574, IEEE, 2013.
COMPARATIVE OBSERVATION OF THE THREE DEVELOPED METHODS AT A
[11] R. Q. Hu, Y. Qian, "An energy efficient and spectrum efficient wireless
NODE DENSITY OF 40
heterogeneous network framework for 5G systems." IEEE
Average Communications Magazine vol. 52.5 (2014), pp. 94-101, 2014.
Method Allocated DR Throughput Delay
Q_Length [12] J. B. Rao, A. O. Fapojuwo, "Analysis of Spectrum Efficiency and Energy
Efficiency of Heterogeneous Wireless Networks With Intra-/Inter-RAT
Multi- Offloading," IEEE Transactions on Vehicular Technology 64.7 (2015),
attribute [73 73 40 20 [4 4 9 pp. 3120-3139, 2015.
[11 11 6 3 2] [0 4 8 17 29]
optimization 13] 12 13] [13] H. Shi, J. Li, Z. Li, H. Ju. "Dynamic spectrum allocation in heterogeneous
[1] wireless networks under interference constraints across the cell
QoS coverage." Wireless Communications and Mobile Computing15, no. 7
Buffer [73 73 47 32 [4 4 7 (2015), pp. 1188-1197, 2015.
[11 11 8 5 5] [0 4 8 15 22] [14] B. Iancu, D. Vasile, A. Peculea. "A novel end-to-end QoS framework over
optimization 27] 10 11]
[14] heterogeneous networks-an architectural approach."Acta Technica
Napocensis 51.2 (2010): 42, 2010.
Proposed [15] L. Badia, M. Miozzo, M. Rossi, M. Zorzi, „Routing schemes in
[73.33 73.33 [4 4 5
approach [11 11 10 9 8] [0 4 8 13 19] heterogeneous wireless networks based on access advertisement and
66 60 53.6] 6 7]
optimization backward utilities for QoS support [Quality of Service based Routing
Algorithms for Heterogeneous Networks],“ . IEEE Communications
Magazine,45(2), pp. 67-73, 2007.
CONCLUSION [16] K. Yang, Y. Wu, H. H. Chen. "QoS-aware routing in emerging
heterogeneous wireless networks [Quality-of-Service-Based Routing
In the suggested approach, the cognitive property of devices Algorithms for Heterogeneous Networks]." IEEE Communications
is utilised for a secondary spectrum sensing operation to Magazine 45.2 (2007), pp. 74-80, 2007.
achieve a higher data rate allocation under ideal spectrum [17] F. Richter, A.. J. Fehske, P. Marsch, G.P. Fettweis, (2010, May). Traffic
availability. In the flow control approach, the buffer control demand and energy efficiency in heterogeneous cellular mobile radio
networks. In Vehicular Technology Conference (VTC 2010-Spring), 2010
mechanism is improved for the given quality of service, where IEEE 71st, pp. 1-6. IEEE.
the traffic monitoring is introduced based on the allocated data [18] G. Fodor, E. Anders, A. Tuoriniemi. "Providing quality of service in
rate and buffer status. An improvement in the network always best connected networks." IEEE Communications Magazine 41.7
operation metric of throughput is observed. (2003), pp. 154-163, 2003.
[19] G. Aristomenopoulos, T. Kastrinogiannis, Z. Li, S. Papavassiliou, (2011),
“An autonomic qos-centric architecture for integrated heterogeneous
REFERENCES wireless networks”, Mobile Networks and Applications, 16(4),pp. 490-
[1] R. Amin, J. Martin, J. Deaton, L. A. DaSilva, A. Hussien, A. Eltawil, 504, 2011.
“Balancing spectral efficiency, energy consumption, and fairness in future [20] J. C. Clement, B. B. Reddy, D. S. Emmanuel, "An Energy-Efficient
heterogeneous wireless systems with reconfigurable devices,” IEEE Cooperative Spectrum Sensing Strategy with Robustness Against Noise
Journal on Selected Areas in Communications, vol. 31(5), pp. 969-980, Uncertainty for Cognitive Radio Networks." Arabian Journal for Science
2013. DOI: 10.1109/JSAC.2013.130515 and Engineering (2015), pp. 1-7, 2015.
[2] S. K. Sharma, S. Chatzinotas, B. Ottersten. "Interference alignment for [21] Z. Dai, J. Liu, C. Wang. "QoS-based device-to-device communication
spectral coexistence of heterogeneous networks," EURASIP Journal on schemes in heterogeneous wireless networks." IET Communications 9.3
Wireless Communications and Networking , vol. 201(1), pp. 1-14, 2013. (2015), pp. 335-341, 2015.
DOI: 10.1186/1687-1499-2013-46 [22] R. Xie, F. R. Yu, H. Ji. "Dynamic resource allocation for heterogeneous
[3] S.V. Hanly, C. Liu, P. Whiting, "Capacity and stable scheduling in services in cognitive radio networks with imperfect channel
heterogeneous wireless networks." IEEE Journal on Selected Areas in sensing." IEEE Transactions on Vehicular Technology 61.2 (2012): 770-
Communications vol. 33(6), pp. 1266-1279, 2015. DOI: 780, 2012.
10.1109/JSAC.2015.2416971 [23] W. Sheikh, A. Ghafoor, “Quality-of-service routing in heterogeneous
[4] H. M. Tahir, A. H. Al-Ghushami, Z. R. Yahya, "Selection of access networks with optimal buffer and bandwidth allocation”, In2007 IEEE
network using cost function method in heterogeneous wireless International Conference on Communications (pp. 480-485) IEEE, 2007.
network." Multimedia Computing and Systems (ICMCS), 2014
International Conference on. IEEE, 2014.
International Journal on Communications Antenna and Propagation (I.Re.C.A.P.), Vol. 7, N. 1
ISSN 2039 – 5086 February 2017
Abstract – The Heterogeneous network has the advantage of utilizing multiple wireless
architectures to exchange information over a wireless medium. In coordination with multiple
devices, long range data transmission is achieved. The advantage of multiple network utilization
has brought out the significance of higher service applications for any network device with the
advantage of using multiple networks for communication. In this network, the demand for proper
resource utilization arises due to a shift to multiple networks. As the network switches from one
network to another, and observed issues like fairness, power, spectrum utilization in the resources
are varied, which impacts the flow of data. Hence it is required to optimize the resource utilization
to achieve a fair and efficient communication. Towards achieving the objective of fairness in an
heterogeneous network with cognitive devices, a multi objective coordination approach for
optimal resource utilization is proposed. The resource utilization problem is defined by an
effective spectrum utilization among the network users. The simulation results show a significant
improvement in resource utilization compared to conventional approaches. Copyright © 2017
Praise Worthy Prize S.r.l. - All rights reserved.
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved DOI: 10.15866/irecap.v7i1.11203
72
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
available either for new services or to expand the existing sensing in this case. This offers a higher throughput
ones. To make sure that existing services can continue to performance in CRN. A decentralized solution where a
grow and evolve to serve market place needs, and that scenario of coexistent heterogeneous CRNs with
new services have a chance to develop, it is important collusion-based PUs is discussed in [11]. A new media
that the Commission continues to improve efficient access control (MAC) protocol for cognitive Wireless
access to and use of the radio spectrum. networks is described in [12] and a distributed multi
The Federal Communications Commission (FCC) channel MAC (MMAC) protocol as enabler for CR
proposed the recommendations in [5] and [6] which networks was presented in [13]. To achieve the detection
allowed the unlicensed device to suitably sense and of free PU spectrum, a collision based approach,
access the licensed spectrum suitably. This applying the concept of Jamming, was suggested. The
recommendation motivated the study of capacity and effect of jamming on the network throughput was
delay scaling laws of cognitive radio networks (CRN) addressed. However, the fairness metric was evaluated in
that considers the coexistence of licensed primary users terms of network throughput and no concern was given
(PUs) and the Unlicensed secondary users (SUs). [7] to the quality of the communicating data. No approach
presents a study on the impact of mobility heterogeneity was suggested to control the data lost in case of observed
of secondary users and static primary users in cognitive distortion due to the propagating channel. The node
radio network. positioning in the utilization of resources for spectrum
In CRN, the spectrum efficiency of the radio unit can sensing or spectrum utilization is also not directed. This
be improved by implementing the method of dynamic paper presents the issue of distortion monitoring and
spectrum access approach (DSA). The users in such quality governance w.r.t. spectrum allocation. The global
network operate as secondary users (SU), which employ issue of node position in spectrum sensing is also
the frequency bands assigned to a primary user (PU). addressed in this paper.
This kind of spectrum utilization is standardized under To present the stated objectives, this paper is
IEEE 802.22 and ECMA-392. organized as follows. Section II outlines the issue of
In the perspective of quality improvement, a joint unfairness in a heterogeneous network. The conventional
optimization approach of energy efficiency and spectrum modelling of decentralized MAC protocol for resource
efficiency for a cognitive wireless network is proposed in allocation is presented in section III. Section IV presents
[8]. The problem of spectrum allocation is defined as a the multi objective coordination proposed approach of
mutual interference model of spectrum efficiency (SE) multi objective coordination approach for co-existing
and energy efficiency (EE) in cognitive radio networks. cognitive heterogeneous networks under quality
In the CRN approach, the detection of Primary user (PU) constraints. In Section V, the proposed solution is
spectrum status is a prime issue. evaluated with simulations and it is shown to improve
The remaining spectrum is utilized by the secondary performance. Finally, conclusions are given in Section
users to communicate and re-allocate to the primary user VI.
when desired. In the real time scenario, two sensing
errors may occur, the misdetection error and the false
alarm error. In the misdetection error the secondary user II. Unfairness Issue in Heterogeneous
detects the primary user spectrum as vacant, when Network [11]
actually engaged. Whereas, in false alarm condition, the
In heterogeneous networks, the existing unfairness is
user detects a spectrum as engaged, when the spectrum is
due to two reasons such as un-coordinated PU detection
free. These two errors minimize the probability of
and uncoordinated spectrum unit size.
detection in CRN. The misdetection error, the result of
co-channel interference and false alarm minimizes
spectrum utilization efficiency. A similar approach in [9], II.1. Uncoordinated PU Detection
an analysis of SE to EE for CRN, was developed. This
approach has investigated the spectral and energy Since PU’s utilization determines a SU’s throughput,
efficiency in interference-tolerant CR networks. This the PU detection ability is important for SU’s to measure
approach defines an additional parameter of link the correct PU utilization in the coexistent CRNs
information to optimize the SE-EE problem. This scenario. Since PU utilization depends on the PU-
approach optimizes the problem of link fading. The detection ability of SUs, the unfairness problem occurs in
primary user channel interference is considered to bound case of uncoordinated PU-detection abilities between
the derived interference margin in allocating the sensed SUs. The uncoordinated PU-detection ability condition
spectrum so that cognitive secondary users (SUs) can occurs in coexistent heterogeneous CRNs cases. Thus,
achieve higher spectrum efficiency while limiting their the iniquity of the uncoordinated PU-detection ability
interference in the PU network. Quality governance in plays a significant role in the assumptive scenario. If it is
this case is observed to be high due to the interference assumed that there are two secondary users such as SU1
limit of primary user for spectrum allocation. In [10], a and SU2 along with a single primary user PU, the
frame structure is proposed for the imperfect spectrum unfairness problem arises if any one of the secondary
prediction. The Ideal state condition is used for spectrum users doesn’t have the PU detection ability. For a given
SU1 and SU2, in absence of PU, the channel will be
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
73
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
occupied by SU1 whereas SU2 also tries to occupy it. based PU-detection mechanism,which is called a
But due to the non-existence of PU detection ability, it “probing function,” is employed to enhance SUs
cannot occupy resulting in unfairness in the throughput abilities.The probing function is a PU-detection
of SU2. In [11], this unfairness problem due to mechanism on the MAC layer without additional
uncoordinated PU detection is solved through a MAC hardware support. It combines with a self-blocking
layer approach which improves the performance of PU mechanism for the uncoordinated spectrum unit size
detection. The main objective of [11] is to identify if the problem.The concept of probing is a trial-and-error
user who occupies a busy channel belongs to a PU or a mechanism that uses jamming to identify the user who
SU from another CRN using a low-overhead mechanism. occupies a channel illustrated in Fig. 1.
It is an assistant function to improve the PU-detection
ability to achieve a better Dynamic Frequency Selection
(DFS) decision.
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
74
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
control the interference level to a PU from jamming, and Here, rmax is the maximum offered data rate and ∆ is
it can improve the PU detection ability via the self- the data rate allocation step related to the jamming rate.
blocking probing function. The key to implement the Once the data exchange is implemented successfully, the
probing function is to find the appropriate p. To deal with performance of the application is not basically affected
the second unfairness problem of uncoordinated by its offered services.
spectrum unit size, p must be modified. However, The backoff delay arises when more than two nodes
modelling an MC that contains multiple-channel and compete for the spectrum resource. This delay will
multiple-system states with different spectrum unit sizes depend on the number of competing nodes on each
is inefficient. Thus, the following equation is used for spectrum band. Thus, during spectrum characterization,
translating jamming rate for simplification purposes: the number of SUs in a distributed CRN is important to
estimate the switching delay.
1 − (1 − ) = (1) Here, the number of packets for transmission is used
to measure the impact of service on QoS. The packet
where indicates the jamming rate of the system with a delay Ta as well as the data delay per packet , can be
spectrum unit size is equal to n. denoted as follows:
= × + (6)
IV. Multi Objective Distribution
Function Approach
= × = × × (7)
Cognitive devices were limited to PU and SU. 1
Recently, it was proposed that the higher spectrum can be
utilized by introducing TSUs from other regions. This
provision builts up the jamming conditions which Here, the parameter c is the proportional coefficient, d
introduces delay in the network. In the proposed is a constant that is determined by network status. ×
approach, the SQL is monitored through delay for the is the time of packet delivery in one unit (how many
effective utilization of the TSU. packets for exchange were sent in one unit). At last, the
If the channel is found to be occupied after conducting end-to-end delay can be obtained by:
a sensing operation, a backoff in time is initiated.
However, if the channel is occupied in repeatedly sensing =( + )/(1 − × ) (8)
operations, then the backoff is also linearly incremented.
On the other hand, if the channel is found to be where, is the network delay and is the processing
‘available’ the next time instance is utilized for delay i.e. encoding and decoding time that will be
transmission and time resolution is backed-off by unity. constant. Then can be controlled by i.e. the
This is a conservative approach since the highest priority spectrum available to a user from its own as well as from
is placed on maintaining the interference to a minimum SU and TSU can not be utilized totally if is
level while trying to improve the other performance increasing due to jamming. To observe the impact of the
metrics. To simplify simulation and analysis, the most exchange process over the transmission operation, an
common service quality measure was applied to our analysis of the data exchange process of transmission and
model. In this case, service quality is measured by the reception is proposed.
number of packets in unit time. In a given network, ra is It is well known that data services can influence SQL
defined as the arrival rate of packets. The value of SQL is metrics, such as end-to-end delay, call dropping
relative to data rate. When the SQL of the packet is not probability and throughput of communication. In this
larger than the offered rate lm, it can be carried out as paper, the end-to-end delay is approximately taken as
follows: SQL because it is the most important among various
SQL factors. The end-to-end delay is defined as the time
= ( − ) (2) for a packet to be sent from one end to the other. If both
encoding and quality encoding are applied, the time
=( − )× (3) delay T can be written as:
Here, la is the SQL of packet with the allocated rate ra, = + + (9)
e is the proportional coefficient of data rate. rmin is the
minimum data rate used in simulations and rm is the In practical application, spectrum sensing is used as a
allocated rate corresponding to lm. When the SQL is general way to protect network resources by rejecting
higher than lm, the additional SQL of the service can be unauthorized spectrum allocation. It is necessary to take
expressed as: over both quality and throughput techniques at relative
high SQL. lm is defined as the minimal SQL where the
=( − )/( − ) (4) encoding algorithm is deployed. The SQL below lm is
mainly determined by encoding, while the additional
∆ =( − )× (5) SQL above lm is obtained from the minimum SQL
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
75
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
between encoding and decoding. This is because both each hope, the node dissipates power based on the IEEE
backoff and encoding can influence the service level, and 802.11 standards for receiving, transmitting, and ideal
weak service from either sides can degrade the SQL conditions. For the developed communication, the
performance. By this means, the final SQL L can be obtained parametric observations are as illustrated in the
expressed as: following figures.
The average throughput for the developed system is
≤ shown in Fig. 3. The average throughput for the proposed
= (10)
+ min(∆ ) > approach of service governed at device nodes results in a
higher throughput as they are communicated with
where, ∆ is the additional increment of rate allocation. minimal delay rate. It was observed that the throughput
for the linear network with energy conservation is also
improved. However, with the increase of number of
V. Experimental Results communication iterations, it was observed that
throughput decrease is due to a congestion of nodes and
For the evaluation of the proposed approach, the
time taken to release the packet. However, the throughput
network parameters used for the simulation are given in
is comparatively observed to improve in case of proposed
Table I.
MDF approach.
TABLE I
NETWORK PARAMETERS USED FOR SIMULATION 140
Linear Network(LT)
Network parameter Characteristic
Random
Node density (Nd) 30,50 120 MAC
15 1200
16 Linear Network(LT)
15 7
27 Random
26
13 1000 MAC
14 MDF(proposed)
10 2
29
11
Network Lifetime (sec)
22 800
28
5 4
3
10
1 600
23
0
0 5 10 15 20 25 30
X-Cordinates 400
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
76
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
It is observed that the network lifetime is increased by simulated network are presented below. For the
the incorporation of energy conservation at node level. simulation, a network with a node density of 50 nodes is
Power at each node is measured and it is observed shown in Fig. 7. The scattering of nodes in such network
from Fig. 5 that, with the increase in the communication can be seen. Due to higher density, the nodes are very
time period, the power level in each node is minimized close to each other. This leads to more route probability
due to energy dissipation during transmission and and more reliability. However, as the number of nodes is
reception operation. However, due to the incorporation of more, the probability of nodes participating in the data
the conservation approach to the developed network, it is forwarding process also increases, resulting in faster
observed that the power level for active nodes is power drain.
increased. This improvement is higher in the proposed The average throughput w.r.t. variation in node
MDF approach. The Network overhead is observed to be density is observed. It is seen from Fig. 8 that the
minimized in case of the topology preserved with energy throughput for the developed approach is improvised
conservation, as shown in Fig. 6. The concept of energy with an increase in node density. The average node
conservation makes more nodes available in the network, density available for the routing in such case increases
which results in higher throughput. Due to more traffic and due to faster processing and rescheduled
clearance the overhead is observed to be less in the conservation, nodes are processed for higher data
proposed approach. transfer. As the data transfer is higher in such network
the observed quality and the network reliability increases.
100
Linear Network(LT)
Randomly distributed Network
Random 50
32 41
95 MAC 2
5 21
MDF(proposed) 45 20
24 49
9 29
90 40 32 23
7 21
15 16 37
10 1 45
Power (mW)
35 29 28
13 3
18 18
85 4417 6 24 40
30 181112 34
Y-Cordinates
1 19 39 22
291 3635 3014
25 2226 7 35
80 40 28 6
38 39
13 47 14 10
33 20
20 1916 2 31
8 16 11 5
27
10 469 37
26 146
75 15 5 25 12 3 9 15 22 25
1 833 4 258 5 4 4
3 11 8 14 21 1711 19 10 26 13
10 8
15
3 5 2 6 28 2 43 42 31
70
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6 5 1 2797 13 30
20
12 24 27
4
16 9 30 3 19
6 18
7 47 17
23 2 50 38
Iterations
0 10 12 15 2023
17 48 36 34
0 5 10 15 20 25 30 35 40 45 50
X-Cordinates
Fig. 5. Power level with communication iteration
9 120
MDF(proposed)
Overhead
8 110
7 100
6 90
5 80
4 70
3 60
1 1.5 2 2.5 3 3.5 4 4.5 5 5.5 6
Iterations 50
40
Fig. 6. Network overhead over communication iteration 10 15 20 25 30 35 40 45 50
Node Density
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
77
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
These features increase the power per node in the improved. In the case of MDF driven power scheduling
network, hence resulting in a longer life time compared approach with energy conservation, the nodes are
to the observation for network life time for fixed node scheduled for sleep and wakeup period, as well the
density, as shown in Fig. 3, this network lifetime gets master node keeps the energy refreshment, which results
increased; due to a large number of nodes, it remains at a in higher power in the network, as observed in Fig. 10.
higher energy level. The Network overhead in such case is observed (in
Fig. 11) to be optimized in case of the topology
800
preserved with energy conservation. The overhead in
Linear Network(LT)
Random such case is reduced, due to a faster data release, due to
700
MAC the availability of more nodes for data exchange
MDF(proposed)
600 compared to its conventional counterparts.
Network Lifetime (sec)
500
VI. Conclusion
400
This paper outlines a Multi Objective distribution
300 Function approach to achieve the objective of fairness in
heterogeneous networks with cognitive radio devices.
200 The network is outlined with different devices of
spectrum sensing with primary users and secondary
100
10 15 20 25 30 35 40 45 50 users. To optimize the free spectrum utilization, a TSU
Node Density
user is defined, which is used to sense free spectrum
from other standing free users from other network
Fig. 9. Network Lifetime over node density
clusters. The objective of spectrum sensing is served via
100 multiple attributes monitoring, and fairness is measured
Linear Network(LT)
Random
in terms of offered service quality with higher
MAC throughput. The offered service quality is observed in
95
MDF(proposed)
terms of service quality level, measured as a parametric
value for the offered service in heterogeneous network.
90 The suggested approach defines the allocation rate
Power (mW)
References
Fig. 10. Power in the network over Node density
[1] Khan, Muhammad Asad, Supeng Leng, Wang Xiang, and Kun
13
Yang. "Architecture of heterogeneous wireless access networks: A
Linear Network(LT) short survey." In TENCON 2015-2015 IEEE Region 10
12
Random
Conference, IEEE, 2015 Pages 1-6.
11 MAC
MDF(proposed)
DOI: 10.1109/TENCON.2015.7372905
10 [2] Yahia Tachwali, Fadi Basma, and Hazem H. Refai, “Cognitive
Radio Architecture for Rapidly Deployable Heterogeneous
9 Wireless Networks” IEEE Transactions on Consumer Electronics,
Overhead
8
Vol. 56, (Issue 3), August 2010.
DOI: 10.1109/TCE.2010.5606279
7 [3] Shao-Yu Lien, Kwang-Cheng Chen, Yonghua Lin,” Cognitive
6
Radio Resource Management For Future Cellular Networks”.
IEEE Wireless Communication. February 2014. Pages 70-79.
5 DOI: 10.1109/MWC.2014.6757899
4
[4] Anusha, M., Srikanth, V., An Efficient Mac Protocol for
Reducing Channel Interference and Access Delay in Cognitive
3 Radio Wireless Mesh Networks, (2016) International Journal on
10 15 20 25 30 35 40 45 50
Node Density Communications Antenna and Propagation (IRECAP), 6 (1), pp.
14-18.
DOI: 10.15866/irecap.v6i1.7891
Fig. 11. Network overhead with variation in Node density [5] Force, FCC Spectrum Policy Task. "Report of the spectrum
efficiency working group." 2002.
The power consumption of such networks will be https://transition.fcc.gov/sptf/files/SEWGFinalReport_1.pdf
lowered and hence the conserved power per node is [6] “Cognitive radio technologies proceedings” Washington, DC,
USA. No. 03-108. https://www.fcc.gov/oet/cognitiveradio
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
78
Syed Shakeel Hashmi, Syed Abdul Sattar, K. Soundararajan
[7] Yi Qin Ying Zhe et al. “ Near optimal scheme for cognitive radio Dr. Syed Abdul Sattar received his B.E Degree
networks with heterogeneous mobile secondary users”. IEEE in Electronics & Telecommunication
Transactions on Communications, vol 63, (Issue 4), April 2015. Engineering from, B.A.M. University
Pages 1106-1120. Aurangabad, India in 1990. He has obtained M.
DOI: 10.1109/TCOMM.2015.2402055 Tech. in DSCE from JNTU Hyderabad, in 2002,
[8] Shaowei Wang, Chonggang Wang., “Joint optimization of and did his first Ph.D. from Golden State
spectrum and energy efficiency in cognitive radio networks”, University, USA, with Computer Science in
Digital Communications and Networks, vol 1,(issue3) August 2004, and a second Ph.D. from JNTU
2015, Pages 161–170. Hyderabad, India with ECE in 2007. He is currently working as Prof &
10.1016/j.dcan.2015.09.004 Director R&D- NSAKCET, Hyderabad- India. His area of
[9] Fourat Haider, Cheng-Xiang Wang, Harald Haas, ErolHepsaydir, specialization is a wireless communications and image Processing.
XiaohuGe, and Dongfeng Yuan, “Spectral and Energy Efficiency
Analysis for Cognitive Radio Networks”, IEEE Transactions on Dr. K. Soundararajan obtained his engineering
Wireless Communications, Vol. 14, (Issue 6), February 2015, degree from SVU Tirupati and Masters from
pages-2969-2980. JNTU and PhD from IIT Roorkee. He is a
DOI: 10.1109/TWC.2015.2398864 distinguished Professor of ECE with Three
[10] Jian Yang, and Hangsheng Zhao, “Enhanced Throughput of decades of teaching and Research experience,
Cognitive Radio Networks by Imperfect Spectrum Prediction”, including two decades of Senior Professor /
IEEE Communications Letters, Vol. 19,( Issue 10), October 2015 Executive, experience as HOD, Principal, and
Pages 1738-1741. Rector at JNTU Anantapur. Currently he is
DOI: 10.1109/LCOMM.2015.2442571 working as Prof & Dean R&D in Teegala Krishna Reddy Engineering
[11] Yu-Chun Cheng, Eric HsiaokuangWu, and Gen-Huey Chen, “A College Hyderabad India.
Decentralized MAC Protocol for Unfairness Problems in
Coexistent Heterogeneous Cognitive Radio Networks
ScenariosWith Collision-Based Primary Users”, IEEE Systems
Journal, VOL. 10,(Issue 1), March 2016. Pages 346-357
DOI: 10.1109/JSYST.2015.2431715
[12] S. Hu, Y.-D. Yao, and Z. Yang, “MAC protocol identification
using support vector machines for cognitive radio networks,”
IEEE Wireless Commun.,vol. 21, (Issue. 1), Feb. 2014 Pages 52–
60.
DOI: 10.1109/MWC.2014.6757897
[13] M. Timmers, S. Pollin, A. Dejonghe, L. Van der Perre, and F.
Catthoor, “A distributed multichannel MAC protocol for multihop
cognitive radio networks,”IEEE Trans. Veh. Technol., vol.
59,(Issue 1), Jan. 2010, Pages 446–459.
DOI: 10.1109/TVT.2009.2029552
[14] C. Luo, F. R. Yu, H. Ji, and V. C. M. Leung, “Cross-layer design
for TCPperformance improvement in cognitive radio networks,”
IEEE Trans. Veh.Technol., vol. 59, (Issue 5), June 2010, Pages
2485–2495.
DOI: 10.1109/TVT.2010.2041802
[15] X. Yiping, R. Chandramouli, S. Mangold, and N. Sai Shankar,
“Dynamic spectrum access in open spectrum wireless networks,”
IEEE J. Sel. AreasCommun., vol. 24,(Issue 3), Mar 2006 Pages
626–637.
DOI: 10.1109/JSAC.2005.862415
[16] Wireless LAN Medium Access Control (MAC) and Physical
Layer (PHY) Specification, IEEE 802.11, 1999. http://www.di-
srv.unisa.it/~vitsca/RC-0809I/IEEE-802-11.pdf
Authors’ information
Syed Shakeel Hashmi received his B.E. Degree
in Electronics & Telecommunication
Engineering from, B.A.M. University
Aurangabad, India in 1999. He has obtained
M.E. in Electronics & Communication
Engineering from, Osmania University,
Hyderabad, India in 2006. He is pursuing PhD
from JNTU Anantapur. He is currently working
as assistant professor in the dept. of Electronics and Communication
Engineering at ITS, ICFAI Foundation for Higher Education (A
Deemed University) Hyderabad India. He is a member of IEEE. His
areas of interest are Wireless Communication, Heterogeneous wireless
networks and signal Processing.
Copyright © 2017 Praise Worthy Prize S.r.l. - All rights reserved Int. Journal on Communications Antenna and Propagation, Vol. 7, N. 1
79
Link Cost Optimization for Traffic Controlling in Heterogeneous
Network
Abstract: Due to the nature of various types of same filed, the large units having high power
end-to-end mobile devices in heterogeneous transmission, and resource utilization are also
network, congestion was one of the main present. Hence, they formulate a heterogeneous
problems in heterogeneous networks. This paper network, which is a integration of multiple
proposes a congestion aware routing approach resources at one area. In such network, the basic
based on the evaluation of weight value as a link problem of communication resources, leads to
cost index. The link cost evaluation considers the heavy traffic congestion, which highly impact the
evaluation of buffer delay, data rate, MAC network performance. The reliability of such
overhead and link quality. This approach uses a network for communication gets minimized. Hence
multipath routing approach which discovers to achieve a reliable communication in such
multiple disjoint routes from a source to networks.
destination having heterogeneity between each This paper proposes a congestion aware routing
other. Among the discovered routes, the route approach by considering the parameters of
with minimum cost index is selected, which is allocated data rate, MAC overhead and delay, by
based on the node weight of all the in-network defining a link quality metric in heterogeneous
nodes from the source node to the destination network. On considering these parameters, a weight
node. By simulation results, the proposed for a given link is evaluated. Using the weight
approach has proved that it attains high packet factor then routes are derived. The route with
delivery ratio, low routing overhead and also lowest minimum cost is considered for
low end-to-end delay. communication. To present the stated approach this
Keywords: Heterogeneous network, congestion, paper is outlined in 6 sections. Wherein section 2
packet delivery ratio, routing overhead, End-to- outlines the conventional approaches of the related
End delay. works. Section 3 present the conventional approach
of congestion aware routing. The proposed
I.INTRODUCTION approach of link weight based routing is defined in
Heterogeneous network are formulated by a set of section 4. Section 5 presents the obtained
nodes, which are dynamic in nature and formulated simulation observations for the proposed work. The
by a group of nodes, communicating over a concluding remarks are presented in section 6.
common wireless medium. Such networks are
characterized by dynamic networks, have variable II. RELATED WORK
link capacity with constraint security and resources.
Routing in Heterogeneous networks is analyzed Congestion in Heterogeneous networks leads to
from a long time. In recent years, various routing transmission Issues, developing delay and packet
protocol for such network have been proposed [1- loss in heterogeneous environment. The issue of
4].In such approaches the method is developed congestion leads to resource wastage in such
under the assumption of nodes with homogeneous network. By the introduction of routing protocols
property. The homogenous property is defined by a which are dynamic with respect to congestion, can
common value been shared among all nodes in the greatly improve the network performance. The
network. increase in data overhead and, due to issues with
However heterogeneous network suffers with multiple devices having variant characteristic, the
network dynamicity, such as, with the variation in performance is degraded. This is predominant in
node density, the performance is highly effected. the case of mobile users in heterogeneous network.
In the applications of such network in real time The mobility concern was referred in past and
application, the nodes are basically observed to various approaches were developed to overcome it.
have heterogeneity property. In various real time In [1] smart personal information system (smart
applications [5-7], the nodes characteristic are PIN), was proposed. This network uses a
variable with respect to their position, and performance over cost to exchange information’s
utilization. For example, in the battlefield area, among users to exchange the data adaptively, using
where the solider carries a portable unit, large units the network characteristic and users requirements.
are carries by heavy mobile units, such as vehicle, In [2] towards a resource monitoring approach, a
tank etc. In such scenario the portable devices are battery lifetime optimization using route selection
low resource in terms of power, processing approaches. In this approach, a vertical handoff
capacity, transmission rate etc. Wherein in the decision was proposed to optimize the offered load
2. Route Reply
The Destination node D sends the route reply
packet RREP along with the total node weight to
the immediate upstream nodeN3.
V. SIMULATION RESULTS
This section gives the details about the
performance evaluation of proposed approach. The
proposed approach is tested for network parameters
of Node mobility speed and packet size. The
proposed WCARP (Weight based congestion aware
routing protocol) is compared with conventional
SCTP [3] and CARM [16] approach.
VI. CONCLUSION