Dynamic Adaptive Resource Allocation Scheme for Multimedia Services in Wireless and Mobile Networks
Summary (12 min read)
1.1.1 Early Wireless and Mobile Networks
- In 1946, the first public mobile telephone network namely Mobile Telephone System (MTS) [4] was introduced in 25 cities in the United States.
- Also, due to the limitation on the number of available channels, only three voice calls could be carried out at the same time in a common service area.
- It also featured full duplex operation and automatic switching of calls thereby eliminating the need for human operators.
- Both the MTS and IMTS systems used a single BS with a high power transmitter to provide service in the entire operating region.
1.1.2 First Generation (1G) Cellular Networks
- A major breakthrough in wireless communications was achieved when researchers working at AT&T Bell Labs in the 1960s, proposed a mobile network based on the cellular concept.
- It proposed replacing the high-coverage areas of legacy systems such as MTS and IMTS with a number of smaller low-coverage areas with non-overlapping boundaries, each with its own BS.
- The total channels available in the operating region were divided among the cells by carefully allocating different channel sets for neighboring cells.
- In other regions of the world, systems based on AMPS architecture were developed and put into operation.
- In Japan, the first analog cellular system, the Nippon Telephone and Telegraph (NTT) system [8] went operational in 1979.
1.1.3 Second Generation (2G) Cellular Networks
- With advances in low-rate digital speech encoding and the increased availability of high density integrated circuits (ICs) in the 1980s, it became possible to migrate towards second generation (2G) digital cellular networks [8].
- Digital encryption of voice data would lead to enhanced security and privacy for customers when compared to the total lack of encryption in 1G networks. (d).
- In 1993, another important 2G system based on CDMA namely IS-95 or cdmaOne [22][23], was standardized and the first commercial systems were introduced in South Korea and Hong Kong in 1995, followed by the United States in 1996 [5].
- The systems based on cdmaOne architecture are backwards compatible with existing AMPS equipment and provide data traffic rates at 4.8 kbps and 14.4 kbps.
1.1.4 Third Generation (3G) Cellular Networks
- The 2G systems have achieved tremendous success since their inception in the early 1990s and continue to enjoy widespread deployment throughout the world.
- Fourth Generation (4G) Cellular Networks, also known as 1.5 Future Trends.
- The existence of several diverse 3G standards limits seamless global roaming between different cellular networks for a mobile user with a single handset.
- In addition, there is a fundamental difference between wireless cellular networks (1G, 2G or 3G) and wireless data networks such as WLANs and PANs.
- Figure 1.1 represents a comprehensive snapshot of the different generations of cellular systems as they have evolved in the past and their expected growth in the coming years.
1.2.1 Existing Technologies for Multimedia Services
- For the 2G cellular system, numerous extension technologies namely HSCSD, GPRS (for GSM networks) and cdmaTwo (for IS-95 networks), are already in place to support medium speed multimedia traffic, offering speeds of up to 173 kbps [38].
- These families of extension technologies for 2G networks are collectively termed as 2.5G standards because of their significantly higher data rates.
- Furthermore, with the current ongoing deployment of 3G standards such as cdma2000, WCDMA and EDGE, users can expect to gain access speeds anywhere from 384 kbps to 2 Mbps [39] along with improved support for multimedia services with adaptive bandwidth requirements (variable transmission rate) [40].
- As faster access speeds are made available to the end user, consumers will want to use their mobile handset for a variety of multimedia services, ranging from simple email browsing, file transfers, web browsing to bandwidth hungry services such as voice conferencing, TV-quality video streaming and broadband Internet.
- In the next section, the authors provide an outline of the different service classes that will be offered by the 3G system to accommodate the growing demands for these new multimedia applications.
1.2.2 Classification of 3G Service Classes
- In general, the services that will be provided by the 3G cellular system can be distinguished into four classes [41]: (a). conversational class; (b). streaming class; (c). interac- tive class and (d). background class.
- The authors can further group these distinct classes into two broad types of traffic [42]: real-time (RT) traffic and non-real-time (NRT) traffic, depending upon how delay sensitive the traffic is.
- The conversational and streaming classes (voice calls, streaming video, etc.) are highly delay sensitive classes and are grouped under RT traffic while the interactive and background classes (email, text messaging, etc.) belong to the lower priority NRT traffic because they have the least delay sensitivity.
- The 3G applications and services are also sometimes classified in increasing order of their bandwidth requirements [43]: (a). voice and audio; (b) wireless messaging; (c). switched data; (d). medium multimedia; (e). high multimedia and (f). interactive high multimedia.
- Specific examples of these services, along with their expected capacity demand, are listed in Table 1.2.
1.3 Organization of the Thesis
- The rest of the thesis is organized as follows – Chapter 2 provides an overview of related work in the areas of resource management and admission control for multiple classes of multimedia traffic in wireless and mobile networks.
- The proposed scheme, DARA, is detailed in Chapter 3 with complete description of the system model, the traffic model and the bandwidth allocation rules which govern their scheme.
- Chapter 4 outlines their simulation model and presents the results of their extensive computer experiments.
2.1 Introduction
- Numerous approaches have been published in recent literature proposing Bandwidth Allocation Algorithms (BAA) for wireless and mobile networks.
- The authors present an overview of some of the proposed BAA schemes which specifically address resource management issues for multiple classes of traffic in a wireless environment.
- The authors discussion of these schemes is structured as follows, the authors first classify the schemes into three categories: (a). traditional schemes; (b). static resource allocation schemes and (c). flexible resource allocation (FRA) schemes.
- Traditional schemes are the simplest in nature and either share the whole system bandwidth among the different classes of traffic (CS –– Complete Sharing) or allocate fixed exclusive bandwidth partitions for the different classes of traffic (FP –– Fixed Partitioning).
- Finally, the authors conclude their discussion with the class of FRA schemes.
2.2.1 Complete Sharing (CS) Scheme
- The CS scheme allows unrestricted access to the system bandwidth by any of the competing classes of traffic.
- The calls are treated on a first come first serve basis, i.e., as long as the required number of Bandwidth Units (BUs) requested by an incoming call are available in the system, the call will be accepted.
- A generic depiction of bandwidth allocation in a CS-based system for K different classes of traffic is shown in Figure 2.1, where indicates the bandwidth available for calls belonging to class i traffic and is the total system bandwidth.
- The reason being that the calls belonging to low bandwidth services, i.e., calls which require fewer BUs, monopolize the system resources while starving the calls belonging to higher bandwidth services, especially true in systems with asymmetric traffic loads.
2.2.1 Fixed Partitioning (FP) Scheme
- In a FP scheme, the total system bandwidth is divided into permanent and static parti- tions among the C K classes of traffic with portion of the bandwidth reserved exclu- sively for calls belonging to class i traffic.
- When a call belonging to class i traffic ar- rives in the system, if the required number of BUs are available in the partition, the call is accepted otherwise the call is rejected.
- When a class i call departs, the released BUs are added back to the partition.
If the bandwidth partitions are sensibly chosen, the FP scheme provides
- Comparable QoS performance for all the classes of traffic at the expense of reduced system resource utilization when compared to the CS scheme.
- An insightful study of both the CS and FP schemes, with extensive analytical and simulation models, is detailed in [46] which substantiate their above mentioned deductions.
As mentioned earlier, static resource allocation schemes differentially treat the multiple
- Classes of traffic based on their varied QoS requirements and the current system state.
- 24 This differentiation is typically implemented by using priority schemes or by using dynamic bandwidth partitions for the different classes of traffic.
- It is to be noted here that this family of schemes does not modify the bandwidth of an ongoing call once it has been admitted into the system.
- Because of their static nature, these schemes are unable to take advantage of the flexible bandwidth requirements of multimedia traffic resulting in a suboptimal system performance.
An example of a static resource allocation scheme is provided in [54], where a priority-
- A bandwidth partition is reserved exclusively for handoff pools and the rest of the system bandwidth is shared among all the different classes of traffic.
- In addition, a threshold-based fixed priority scheme is established between the different classes of traffic, which “admits highpriority traffic preferentially under heavy traffic loads”.
- Under this scheme, calls belonging to the low priority traffic will suffer from high blocking probabilities.
A CS-based Dual Threshold Bandwidth Reservation (DTBR) scheme is proposed in [44]
- Which handles two broad classes of traffic –– voice traffic (high priority) and data traffic (low priority).
- The total system bandwidth C is divided into three partitions by using two fixed thresholds and .
- When the total occupied bandwidth is less than , both voice and data traffic can be serviced by the system.
- The bandwidth partitions for the DTBR scheme are shown in Figure 2.3.
- The drawback with 1T 2T )( 12 TT < 2T 2T 1T 1T 25 this scheme is that it is static in nature with fixed partitions; dynamic bandwidth allocation schemes can always achieve better performance than the proposed DTBR scheme.
2.4 Flexible Resource Allocation (FRA) Schemes
- Due to the adaptive nature of multimedia services, most of the multimedia traffic can be encoded at variables rates, i.e., multimedia traffic can occupy multiple levels of bandwidth depending on the current system load and the bandwidth available for transmission.
- In recent literature, numerous FRA schemes for wireless and mobile networks have been proposed which take advantage of these adaptive bandwidth requirements of the newer multimedia services.
A graceful bandwidth degradation strategy for both real-time and non-real-time multime-
- To support increased traffic in the system, some of the ongoing calls are degraded to accommodate more new calls.
- The degradation strategy and call admission control in this scheme is governed by a cost function based on the net revenue generated by the system.
- In [47], the system bandwidth is divided among the different classes of traffic proportional to their arrival rates (inter-class fairness).
- The dynamic resource allocation scheme of [49] modifies the bandwidth of both ongoing and incoming calls according to the current occupied bandwidth in the system.
- These schemes are also considered in their comparisons because they represent the upper and lower performance bounds for any FRA scheme [53].
In this chapter, some of the key topics are presented which provide a better understanding
- Of the proposed scheme, DARA, in subsequent chapters.
- In the first section, the authors introduce three important probability distributions –– exponential, Poisson and Pareto distributions with emphasis on their unique properties which are useful for the modeling of different classes of traffic in a wireless and mobile system.
- Next, the handoff process is described in detail with information about the different types of handoffs, the handoff criteria and the various handoff schemes.
3.2 Important Probability Distributions
- The performance evaluation of wireless and mobile networks is specified by various nondeterministic factors which help define the whole system.
- Some examples of such random parameters are the nature of call arrivals, the expected duration of calls and the movement of non-stationary users in the system [4].
- For the efficient modeling and analysis of any wireless and mobile network, it is necessary to quantify or approximate 29 the aforementioned parameters into well-defined probability distributions.
- Since the underlying probability and statistics theory is too extensive to be covered by the scope of this thesis, the authors briefly define the some of the probability distributions and their important properties that are used to define their system model and traffic model in subsequent chapters.
3.2.1 Exponential Distribution
- The exponential distribution, due to its simplicity, is one of the most widely used distributions in the simulation of computer systems and networks.
- The exponential distribution is the only continuous time random distribution that has no memory, i.e., the probability of an event happening in a specific time interval is the same regardless of the starting point of that time interval.
- In other words, the equipment does not remember that it has already been used for time t , hence the “memory-less” property.
- The exponential distribution is widely used to model the duration of calls and the channel holding time in wireless and mobile systems.
3.2.2 Poisson Distribution
- The Poisson distribution is typically used to model random events occurring over a fixed interval of time.
- This property proves very convenient in simulation of Poisson events in computer programs by letting the inter-arrival times between the events follow an exponential distribution.
- This property can be used to model the combined traffic load on a wireless system serving multiple classes of traffic, with each class being represented by its corresponding Poisson process.
3.2.3 Pareto Distribution
- The Pareto distribution is one of the simplest examples from the family of heavy tailed distributions.
- When the condition 21 << α is true, the Pareto distribution has an infinite variance, a finite mean and exhibits self-similar behavior [57].
3.3 Handoff Process
- The wireless and mobile networks are characterized by one unique factor, i.e., the users of these networks are mobile.
- To provide roaming access to these mobile users as they move across the different service regions, it is necessary to design an efficient handoff algorithm.
- For cellular networks, handoff (also sometimes known as hand-off or handover) can be defined as “the process of changing the channel (frequency, time slot, spreading code or a combination of them) associated with current connection while the call is in progress” [58].
3.3.1 Types of Handoff
- Handoffs can be broadly divided into two categories –– hard handoffs and soft handoffs [4].
- Hard handoffs are characterized by a break before make connection, i.e., the resources held by a Mobile Station (MS) with the current BS (Base Station) are released first before acquiring new resources from the next BS.
- The FDMA and TDMA systems employ hard handoff while soft handoffs are implemented in CDMAbased systems.
- 34 Handoffs are also sometimes classified as intra-cell and inter-cell handoffs [59].
- Intra-cell handoff occurs when there is a channel change within the same cell while in an inter-cell handoff, the channel change occurs from one cell to another cell.
To make a handoff decision, it is necessary to choose certain criteria which provide use-
- Ful information about the call and their threshold values which determine the need for 35 handoff.
- Some of the criteria that are traditionally considered in a wireless and mobile network for handoff decisions are presented below [59].
- But, in practical scenarios, if the authors consider only this parameter, it gives rise to unnecessary frequent handoffs due to rapid fluctuations in the signal strengths of adjacent BSs in the handoff area.
- It is recommended to use the RSS value along with other measurement parameters for a more efficient handoff process.
- It determines the size of the cluster, cell shape and the frequency reuse distance in cellular networks.
In these cases, handoff is done for reasons other than degradation of the channel or the
- Loss of signal strength; network criteria are usually managed by a central network entity such as the Mobile Switching Center (MSC).
- In a directional antenna based system, the channel of the MS might have to be changed when it moves from one sector to another sector 36.
3.3.3 Handoff Schemes
- Once the handoff criteria have been chosen, it is necessary to designate the network entities that are responsible for measuring these criteria and making a handoff decision based on their threshold values.
- Even though the delay associated with just the handoff process is around 100-200 ms, the overall delay can be in the range of 5-10s.
- In a MAHO system, the responsibility for measuring the handoff criteria and making the handoff decision is assigned to the BS and the MSC, respectively.
- An example of a MCHO based handoff control network is the standard for cordless phones in Europe –– DECT (Digital European Cordless Telephone) [61].
- 38 Chapter 4 Dynamic Adaptive Resource Allocation (DARA) Scheme.
4.1 Introduction
- The authors present their proposed BAA (Bandwidth Allocation Algorithm) approach, namely Dynamic Adaptive Resource Allocation (DARA) scheme [50], which is specially suited to handle multiple classes of multimedia traffic in wireless and mobile networks.
- The proposed scheme, DARA, is based on the concept that multimedia traffic has flexible bandwidth requirements and the system capacity can be increased if the authors intelligently degrade some ongoing calls to accept more calls in the system.
- This drastically reduces the QoS fluctuation experienced by a call during its lifetime while giving better system performance in terms of the blocking probabilities for both originating and handoff calls.
- The rest of the chapter is organized as follows: the authors start with a description of the underlying system model and explain the various associated system level parameters.
- The traffic model is presented next and it defines the behavior of call arrivals for the different classes of traffic and channel occupancy by those calls in the system.
4.2.1 System Parameters
- BUs required by a class i call; can also take only integer values.
- After an overview of the system parameters, the authors now proceed to describe their system model in detail.
- In addition, handoff calls can also access the originating bandwidth pool along with the originating calls.
4.2.2 Traffic Model
- The authors describe the traffic model assumed in their scheme for a single cell belonging to a wireless and mobile network.
- In another approach, the Pareto distribution is utilized to generate the inter-arrival times for originating and handoff calls belonging to class i with mean arrival rates of oiλ and hiλ respectively.
- One simple mobility model, which describes the movement of mobile users in cellular system, is proposed in [64] based on the fluid flow model.
- With the definition of their system model complete, the authors now proceed to outline the rules governing resource allocation in their proposed scheme.
4.3 DARA Bandwidth Allocation Algorithm
- In their scheme, depending upon the type of calls (originating or handoff) and the class i of traffic, appropriate set of bandwidth allocation rules are processed.
- All the different call arrival and departure events along with their associated processing rules are defined in the following subsections.
4.3.1 Originating Call Arrival Handling
- If band- width is not available, bandwidth is allotted following the same procedure as above.
- If neither nor bandwidth is available, the incoming call is rejected.
- The detailed steps for originating call arrival handling are given in Algorithm I.
- The bandwidth upgrade/degrade process and the criteria for choosing ongoing calls for the upgrade/degrade procedure are explained in the later sections.
4.3.2. Handoff Call Arrival Handling
- In their proposed scheme, handoff call arrivals for all classes are treated according to identical rules.
- Since handoff calls have the strictest QoS requirements, the system reserves some bandwidth exclusively for the handoff calls.
- The detailed steps for handoff call arrival handling are presented in Algorithm II and the flowchart for both originating and handoff call arrival handling is shown in Figure 4.2.
- IF YES, ASSIGN BANDWIDTH TO INCOMING CALL AND ACCEPT CALL, STOP.
- Minib hC minib STEP 3. IF BANDWIDTH IS AVAILABLE IN THE ORIGINATING BANDWIDTH POOL , AS- SIGN BANDWIDTH TO INCOMING CALL AND ACCEPT CALL, STOP.
4.3.3 Call Departure Handling
- If the departing call was initially generated in the same cell, the released bandwidth by that call is distributed back to the current ongoing calls which are receiving degraded service in the system.
- The bandwidth upgrade process is described in the later sections.
- If excess bandwidth is still left over after upgrading all the calls (possible under low traffic conditions), it is added to the originating bandwidth pool .
- If bandwidth for the departing call was obtained from the originating bandwidth pool or by degrading ongoing calls, the re- leased bandwidth by that call is distributed back to the current ongoing calls which are receiving degraded service in the system.
- The detailed algorithm and flowchart for call departure handling are shown in Algorithm III and Figure 4.3, respectively.
4.3.4 Bandwidth Degrade Process
- The following guidelines are used to borrow bandwidth from ongoing calls when there is not enough bandwidth available in the system to accommodate incoming calls: (a).
- Initially, ongoing calls belonging to the lowest priority class, i.e., the traffic class with the highest are selected.
- Calls across the same class i occupying the highest bandwidth levels are the first ones to be degraded. (c).
- This approach results in freeing up maximum bandwidth for incoming call (either originating call or handoff call) while affecting the least number of ongoing calls thereby reducing QoS fluctuation across the system.
- Calls in a class i already occupying the minimum bandwidth cannot be de- graded further minib.
4.3.5 Bandwidth Upgrade Process
- When a call departure occurs, the released bandwidth is redistributed in the system according to the following guidelines: (a).
- The calls belonging to the highest priority class, i.e., the traffic class with the lowest are upgraded first.
- Calls across the same class i occupying the lowest bandwidth levels are the first ones to be upgraded.
- When an ongoing class i call is upgraded, its bandwidth is increased from to .
- Calls belonging to class i already occupying the maximum bandwidth can- not be upgraded further.
4.4 QoS Metrics
- In the performance evaluation of the DARA scheme, the authors take into account conventional QoS parameters such as the blocking probabilities of originating calls and handoff calls.
- Apart from evaluating the traditional performance parameters, the authors also need to measure the signaling overhead required at the BS for the bandwidth upgrade/degrade process and the proportion of calls receiving degraded service in the system.
- To accommodate these new features, the authors introduce two additional QoS metrics which reflect the overhead involved in implementing a dynamic resource allocation scheme as opposed to traditional static schemes such as CS and FP.
4.4.1 Average Degraded Bandwidth (ADB)
- The value of ADB can vary from to as the amount of traffic in the sys- tem increases.
- This parameter gives an approximate measure of user satisfaction because it indicates the average level of degraded service that a user is subjected to in their system.
4.4.2 Bandwidth Reallocation Frequency (BRF)
- The parameter BRF is defined as the number of times a call undergoes bandwidth reallocation during its lifetime.
- In the next chapter, the authors present the performance analysis of their proposed scheme, DARA, which is accomplished by extensive computer simulations along with detailed discussions of the obtained results.
5.1 Introduction
- For the performance evaluation of their proposed scheme, DARA, the authors adopt an eventdriven simulation model.
- The rest of the chapter is organized as follows:–– the simulation model along with the simulation parameters is described in the first part and the simulation results are presented in the next part.
- In their experiments, the authors consider three classes of multimedia traffic with flexible bandwidth requirements comprising of both originating and handoff calls.
- The HFRA scheme is a dynamic FRA scheme which modifies the bandwidth of both ongoing and incoming calls according to system bandwidth; it serves as the base scheme whose performance the authors aim to improve in their work.
- In addition, the authors also evaluate the performance based on two non-traditional QoS parameters –– ADB (Average Degraded Bandwidth) and BRF (Bandwidth Reallocation Frequency).
5.2.1 Programming Environment
- The programming environment for their simulation experiments is the general purpose programming language, JAVA (build jdk1.5.0) [65], augmented with a discrete-event process oriented simulation package, SimJAVA [66].
- SimJAVA is an API (Application Peripheral Interface) which extends the JAVA programming language to conveniently model and execute event-driven simulations.
- These unique features translate to ease of use for developers and considerable reduction in programming time when used for modeling the random call arrival/departure events in a wireless and mobile network.
5.2.2 Simulation Parameters
- In their simulation model, the authors focus on a single cell of a mobile and wireless network serving three classes of multimedia traffic with handling for both originating and )3( =K 58 handoff calls.
- Since their proposed scheme, DARA, is specifically tailored for multimedia or non-voice traffic with flexible bandwidth requirements, the authors ignore the conversational class which consists of voice traffic.
- Additionally, multimedia services are generally divided into three broad groups based on their bandwidth requirements [67]: 1. Light multimedia traffic such as E-mails, web browsing and so on.
- The arrival processes for originating and handoff calls belonging to class are assumed to be Poisson processes with exponentially distributed channel holding times.
- The rest of the values for the various simulation parameters are listed down in Table 5.1. i 60.
5.3 Simulation Results and Discussions
- The authors present the simulation results obtained by comparing the different resource allocation schemes –– their proposed scheme DARA, the basic dynamic FRA scheme HFRA and the boundary FRA schemes, HMax and HMin.
- The system parameters that are chosen for the performance evaluation are the blocking probabilities 61 of originating calls and handoff calls, BRF (Bandwidth Reallocation Frequency) and ADB (Average Degraded Bandwidth).
5.3.1 Simulation Set 1 (SS1) Results
- For the simulation set 1 (SS1), the inter-arrival times for all the three classes ( K = 3) of traffic are generated according to the Pareto distribution with the shape parameter arrα fixed at 1.50.
- This is a i maxib 63 sub optimal implementation because if maxib bandwidth is not available, minib bandwidth might still be available and can be allotted to the incoming call to increase the system capacity.
- In Figure 5.4, Figure 5.5 and Figure 5.6, the ADB (Average Degraded Bandwidth) for calls belonging to class 1 traffic, class 2 traffic and class 3 traffic are plotted against their respective traffic arrival rates.
- It should be noted that the HFRA scheme does perform marginally better than the proposed DARA scheme.
5.3.2 Simulation Set 2 (SS2) Results
- The authors adopt the traditional traffic model historically used to model the nature of call arrivals and channel holding times in a wireless and a mobile system.
- The call arrival processes for all the three classes of traffic are modeled as Poisson processes while the channel holding times for these calls are exponentially distributed with a mean value of iµ/1 = 120s.
- The proposed scheme, DARA, outperforms the HFRA scheme with substantially lower blocking probabilities for all the classes of traffic.
- I maxib minib 70 Figure 5.16, Figure 5.17 and Figure 5.18 plot the average BRF (Bandwidth Reallocation Frequency) experienced by calls belonging to class 1 traffic, class 2 traffic and class 3 traffic respectively against their traffic arrival rates.
- The observed results are comparable to the results obtained for the simulation set (SS1), i.e., the proposed DARA scheme achieves a significant reduction in average BRF when compared to the HFRA scheme.
5.3.3 Simulation Set 3 (SS3) Results
- For the simulation set 3 (SS3), both the inter-arrival times and the channel holding times for calls belonging to all the three classes of traffic follow Pareto distributions.
- To generate the inter-arrival times of all the three classes of traffic, the shape parameter for the arrival Pareto distribution arrα is chosen to be 1.50.
- For the sake of completeness, the obtained simulation figures are included below .
- To interpretation these results, the reader is directed to the detailed discussions presented for simulation set 1 (SS1) results in the preceding sections as they hold true for this set too.
5.3.4 Simulation Set 4 (SS4) Results
- The simulation results obtained for the different QoS metrics, namely –– blocking probabilities of originating and handoff calls, BRF (bandwidth reallocation frequency) and ADB (Average Degraded Bandwidth), are presented in Figure 5.28 through Figure 5.36.
- The observed results are similar to the results obtained for simulations sets - SS1 and SS2.
- For further explanation of these results, please refer to discussions presented for simulation set 1 (SS1) results in the preceding sections as they hold true for this set too.
5.3.5 Simulation Set 5 (SS5) Results
- For the traffic model, the authors utilize the Pareto distribution to generate call arrivals and the exponential distribution to model the channel holding times.
- From the figures, it is clear that for all the three different classes of traffic, the blocking probabilities become higher as arrα is decreased from 1.8 to 1.2.
- This is understandable because as arrα becomes closer to 1, the traffic becomes more ‘bursty’ and it is quite possible that more calls are lost due to tighter arrivals to the system.
- Because of the non-bursty nature of a Poisson arrival process, when the call arrivals are modeled as a Poisson process (indicated by the ‘Poisson’ plot), the value of the ADB is higher than when the inter-arrival times are modeled as a Pareto distribution (indicated by the ‘alpha 1.2’, ‘alpha 1.5’ and ‘alpha 1.8’ plots).
6.1 Conclusions
- The authors have proposed and analyzed a resource management approach, namely Dynamic Adaptive Resource Allocation (DARA) scheme, which exploits the flexible bandwidth requirements of multimedia traffic to increase system capacity of wireless and mobile networks.
- To address these issues, numerous resource management techniques have been published in recent literatures which handle multiple classes of traffic.
- In the performance analysis of their proposed scheme, the extensive simulation results indicated that the proposed scheme provided enhanced system capacity by allowing incoming calls to borrow bandwidth from ongoing calls at heavy traffic loads.
- In addition, by introducing a dynamic bandwidth borrow unit , the QoS fluctuation or the BRF (Bandwidth Reallocation Frequency) experienced by a user was considerably reduced when compared to a similar FRA scheme [49]. ib∆.
Did you find this useful? Give us your feedback
Citations
3 citations
Cites methods from "Dynamic Adaptive Resource Allocatio..."
...So far, many CAC algorithms have been presented to guarantee QoS of satellite networks [5], [6], [7], [8]....
[...]
2 citations
Cites background or methods from "Dynamic Adaptive Resource Allocatio..."
...In summary, if the classification presented in the previous section is used, it can be observed that the cross-layer schemes in [13-18] used a MAC-centric approach, those in [19-21] used an application-centric approach while the schemes in [2-12] used the integrated approach to combine various layers to guarantee QoS targets while meeting some particular objective....
[...]
...Crosslayer design has been used to design resource allocation algorithms for multimedia QoS guarantees in [2-12]....
[...]
...However, most of these schemes [2-18] only focus on interactions between lower (physical and data link) layers and ignore the possibility of sharing information with, and thus combining, higher layers in the optimization process....
[...]
1 citations
1 citations
Cites methods from "Dynamic Adaptive Resource Allocatio..."
...The RSVP uses the integrated services approach [18-20]....
[...]
Cites background from "Dynamic Adaptive Resource Allocatio..."
...Finally, the conclusions are drawn....
[...]
...To guarantee the users’ QoS in LEO satellite networks, it is necessary that an efficient and reliable resource reservation strategy and call admission control (CAC) algorithm should be designed....
[...]
References
5,567 citations
4,028 citations
"Dynamic Adaptive Resource Allocatio..." refers background in this paper
...If we interpret x as the lifetime of an equipment [56], then (3....
[...]
3,915 citations
2,360 citations
"Dynamic Adaptive Resource Allocatio..." refers background in this paper
...In general, the services that will be provided by the 3G cellular system can be distinguished into four classes [3]: (a)....
[...]
667 citations
"Dynamic Adaptive Resource Allocatio..." refers background in this paper
...[3] H. Holma and A. Toskala, WCDMA for UMTS, radio access for third generation mobile communications, John Wiley & Sons Ltd., Chicester, England, 2000....
[...]
..., n) then the resulting i th process will also be a Poisson process with a mean rate of i P , 1 = i λ i P events per unit time [4]....
[...]
...Some examples of such random parameters are the nature of call arrivals, the expected duration of calls and the movement of non-stationary users in the system [4]....
[...]
...1 Types of Handoff Handoffs can be broadly divided into two categories –– hard handoffs and soft handoffs [4]....
[...]
...Inter-arrival Times of a Poisson Process Starting with an arbitrary time t , let denote the time until the first event occurrence in a Poisson process, then it can be shown [4] that has an exponential distribution with mean rate 1 T...
[...]
Related Papers (5)
Frequently Asked Questions (3)
Q2. What have the authors stated for future works in "Dynamic adaptive resource allocation scheme for multimedia services in wireless and mobile networks" ?
In this section, the authors introduce some of the prospective future issues in their research work.
Q3. What is the purpose of this thesis?
From the extensive simulation results, it is evident that their scheme achieves better performance with regard to traditional Quality of Service (QoS) parameters such as the call blocking probabilities of originating and handoff calls while significantly reducing the bandwidth reallocation frequency experienced by a call when compared to similar flexible resource allocation schemes.