Efficient Resource Management in Future Heterogeneous Wireless ...

The 2010 Military Communications Conference - Unclassified Program - Networking Protocols and Performance Track

Efficient Resource Management in Future Heterogeneous Wireless Networks: the RIWCoS Approach Vladimir Atanasovski, Valentin Rakovic, Liljana Gavrilovska Faculty of Electrical Engineering and Information Technologies Ss Cyril and Methodius University in Skopje Skopje, Macedonia {vladimir; valentin; liljana}@feit.ukim.edu.mk Abstract—Convergence of different wireless access networks and development towards 4G yields a necessity for more efficient resource management procedures in the integrated heterogeneous platform. The notion of resource is fundamentally different in heterogeneous environments (compared to homogeneous ones) as it is spanned across several different wireless solutions. This paper summarizes the development process of a novel architecture for resource management in heterogeneous wireless networks named RIWCoS. The architectural focal point is on the emerging IEEE 802.21 standard because of its ability to enable seamless and transparent vertical handovers. The paper shows simulation results of the IEEE 802.21 benefits and reports on a novel simulation analysis (using newly introduced metrics) of the RIWCoS architecture showing that it outperforms traditional serving policies of users in heterogeneous wireless environments. Moreover, the paper elaborates on a distinguished RIWCoS feature (the SAES algorithm) that enables efficient handling of emergency situations making the RIWCoS architecture suitable for military communications. Keywords- Heterogeneous wireless networks; IEEE 802.21; Resource management; RIWCoS; SAES

I.

INTRODUCTION

Heterogeneous Wireless Networks (HWNs) represent an interoperable set of different Radio Access Technologies (RATs) that offers user transparent and seamless handovers within [1]. The HWN functionalities enable the wireless networks development towards the 4G paradigm emphasizing the user centricity and service personalization. HWNs are also very attractive for military communications being able to provide various forms of (often overlapping) network connectivity. This feature allows them to target the military environment more efficiently provisioning constant users’ connectivity, regardless of the environmental context of the users, managing potential emergency/disaster situations by offering alternative access and offering higher service retainability and accessibility. One of the cornerstone aspects of HWNs is the Resource Management (RM) [2], which is fundamentally different from

978-1-4244-8179-8/10/$26.00 ©2010 IEEE

homogeneous wireless systems. Namely, the aspect of resource in HWNs is spanned across different wireless networks and becomes difficult to manage in a way that guarantees user satisfaction and maximum network utilization. Additionally, HWNs impose “soft” RM (i.e. the RATs may negotiate the dedicated bit rates for their users) introducing the notion of reconfigurability in an HWN context [3]. There are several proposed RM frameworks in the literature today, specifically tailored for the HWN paradigm [2]. For instance, the Multiaccess Radio RM (MRRM) supports network-centralized RM functionalities by a hierarchically higher node than the RATs’ RM entities or peer-to-peer communication among them. The Common Radio RM (CRRM) requires a centralized RM approach that sees a pool of resources from the constituent RATs which are then commonly managed. The Joint Radio RM (JRRM) framework envisions service-split functionalities and multi-homing. The Cognitive RM (CRM) is a novel RM paradigm envisioning artificial intelligence and machine learning mechanisms incorporated into the RM blocks in the network. This paper reports on a simulation analysis of the emerging IEEE 802.21 standard [4] which is then used as a basis for the definition of a novel RM architecture for HWNs. The architecture, named RIWCoS, is developed through a NATO Science for Peace (SfP) funded research activity [5] in the period October 2007 – June 2010. Various stages of the architecture development were previously reported [6-11] while this paper provides a comprehensive overview of the most prominent architectural features, a novel simulation based evaluation of the integrated RIWCoS concept for enhanced RM (using newly defined performance metrics) and architectural upgrades for emergency situations. The RIWCoS architecture itself embraces several concepts specifically modeled for HWNs, exhibits high users’ satisfaction and service retainability and is able to handle disaster situations in an efficient manner. The last feature is especially important for military communications. The paper is organized as follows. Section II will report on some related work in the field of IEEE 802.21 and HWN RM architectures. Section III provides details along with

1592

performance analysis of the IEEE 802.21 benefits. Section IV defines the RIWCoS architecture, whereas section V gives a performance analysis of its ability to retain services in a heterogeneous context. Section VI introduces a specific RIWCoS feature able to cope with emergency/disaster situations. Finally, section VII draws some important conclusions and pinpoints possible future work in the field. II.

RELATED WORK

As previously discussed, the RIWCoS’ architecture focal point is the IEEE 802.21 standard that enables media independent vertical handovers. There are several ideas proposing the standard’s usage by existing mobility and QoS management protocols as a solution to the RM problem in HWN context. For example, the authors in [12] propose the convergence of SIP and IEEE 802.21 as a powerful tool for soft vertical handover execution. Ref. [13] argues that IEEE 802.21 can be used for integration of multimedia broadcast technologies (DVB-H) with other terrestrial access networks. Ref. [8] and [14] utilize IEEE 802.21 for QoS provisioning in IEEE 802.16 – IEEE 802.11 environment. These works show that the assistance of IEEE 802.21 contributes in decreasing the effects of handover latency, jitter and packet loss, thus improving the user perception. Additionally, IEEE 802.21 can be used in a hybrid Satellite – Terrestrial access network [15] or enable Knowledge Based (KB) mechanism for network selection [16]. The RM plays a crucial role in the HWN platform. However, there is a fairly small amount of research work [1720] that specifically addresses this problem. The authors in [17] define a specific approach (based on the JRRM framework) for beyond 3G wireless heterogeneous systems capable of adapting to the resource assignments of the specific system conditions and service QoS demands. Ref. [18] uses a dynamic spectrum management scheme in the JRRM framework in order to increase spectrum utility and reduce blocking rate. Ref. [19] proposes a cooperative RM with policy-based network resource allocation algorithm. Ref. [20] introduces a combination of multiple cognitive algorithms (e.g. genetic algorithms, fuzzy logic, multiple criteria decision making) that results in a single and most optimal decision in the RM process. Unlike previous work in the field, the RIWCoS approach presented here enables a fully functional RM mechanism for HWNs and possesses unique features allowing maximum user servicing, maximum network utilization and efficient coping with emergency/disaster situations. Up to the best of authors’ knowledge, there is no fully functional architecture in the literature yet that relies on the IEEE 802.21 standard in order to enable RM in HWNs. The RIWCoS approach is the first comprehensive example of such integration. III.

is reflected in the minimization of the vertical handover latency value that will facilitate more efficient RM mechanisms. The heart of the IEEE 802.21 framework is the Media Independent Handover Function (MIHF). The MIHF has to be implemented in every IEEE 802.21 compatible device and is responsible for communication with different terminals, networks and remote MIHFs providing abstract services to the higher layers using a unified interface (L2.5 functionalities). MIHF defines three different services, i.e. Media Independent Event Service (MIES), Media Independent Command Service (MICS) and Media Independent Information Service (MIIS). MIES provides events triggered by changes in the link characteristic and status. MICS provides the upper layers necessary commands to manage and control the link behavior to accomplish handover functions. MIIS provides information about the neighboring networks and their capabilities. Extensive details on the IEEE 802.21 standard can be found in [4] and are omitted here due to space limitations. The benefits of the IEEE 802.21 standard are analyzed through extensive simulations in QualNet [21], a commercially available network simulator. The IEEE 802.21 functionalities were implemented in a separate and standalone executable application that communicates with QualNet whenever a vertical handover situation occurs. Details on the complete simulation environment specifically designed for the RIWCoS architecture can be found in [22]. Targeted simulation scenario for IEEE 802.21 performance analysis is depicted on Fig. 1. It comprises an HWN consisting of two IEEE 802.11 Access Points (APs) providing local coverage, two IEEE 802.16 Base Stations (BSs) providing metropolitan coverage and a satellite network for global scenario coverage. All RATs overlap in order to enable increased connectivity options within the HWN. There will be a varying number of mobile nodes in the scenario communicating with Correspondent Nodes (CNs) located in the network infrastructure. The users are allowed to move according to the random waypoint mobility model and they exhibit frequent vertical handovers in the scenario, i.e. frequent change of the Point of Attachment (PoA) serving the users. All mobile nodes have active CBR/UDP applications with varying bit rate for different simulations. The simulations last for 3 minutes allowing the mobile nodes to perform high number of vertical handovers. As all subsequent results will be averaged per user and per scenario, this simulation model increases the confidentiality level of the results. The confidence interval of the simulations is around 95%.

BENEFITS OF IEEE 802.21

Before the RIWCoS architecture is defined and evaluated, this section will focus on the justifications of the usage of IEEE 802.21 standard as a basis for RM architecture building. The main research driving factor in HWNs is the seamless vertical handover performance among different wireless networks. This

1593

Figure 1. Simulation scenario for IEEE 802.21 performance analysis

Main performance parameter of interest from the IEEE 802.21 implementation that is crucial for the subsequent architectural design is the Vertical Handover Latency (VHL). The minimization of the VHL value allows seamless vertical handovers in HWNs and facilitates the RM problem substantially. Its calculation must take into consideration the end-to-end delay of the new serving network and the receiver packet latency once a vertical handover occurs. The former parameter is calculated as: i i - t sent t ei 2 e = t rcv

Fig. 4 depicts a comparison of the average VHL latency for different application bit rates, low mobility and high number of nodes. The introduction of IEEE 802.21 exhibits higher gains (in terms of lower VHL values) for lower application data rates. However, the IEEE 802.21 allows having smaller VHL value even for higher application data rates compared to the case without IEEE 802.21 and lower application data rates.

(1)

i i where t rcv is the packet receive time and t sent is the packet sent time on the new service network (denoted as network i) when a vertical handover happens. Using eq. (1), the VHL value can be easily derived as: ij t VHL = t Lij - t ei 2 e

(2)

where t Lij is the receiver packet latency (defined as the time difference between the last successfully received packet from the old service network j and the first successfully received packet from the new service network i). Fig. 2 and 3 depict the average VHL value for 30 and 90 mobile nodes in the scenario, respectively. It is evident that the introduction of the IEEE 802.21 functionalities exhibits significantly lower values for the VHL (the decrease is 3 to 6 times depending on the simulation scenario). Furthermore, the results show that the IEEE 802.21 allows smaller increase of the VHL value with the increase of the number of mobile nodes (especially for high mobility). Finally, it is clear that the VHL latency with IEEE 802.21 employed is almost independent from the users’ mobility pattern.

Figure 2. VHO latency values for 30 mobile nodes in the simulation scenario

Figure 3. VHO latency values for 90 mobile nodes in the simulation scenario

Figure 4. VHO latency values comparison for different application data rates (90 mobile nodes in the simulation scenario)

The simulation results clearly show that the IEEE 802.21 standard is an excellent reference for enabling seamless vertical handovers. As a result, it is customized and used to propose and define a novel RM architecture, i.e. the RIWCoS architecture, in the following section. IV.

THE RIWCOS ARCHITECTURE

The RIWCoS approach to RM relies on the notion of MIH User defined within the IEEE 802.21 standard. The MIH User uses the services of the lower MIH Function (MIHF) in a standardized way (Fig. 5). In order to encompass all possible user scenarios envisioned in the RIWCoS paradigm, the RM approach requires both network and terminal side RM modules. However, the implementation of network RM modules is more intriguing and requires many assumptions and prerequisites from the network side. Therefore, RIWCoS at the current stage defines and evaluates only the terminal side RM module, Fig. 6. The Application block presents the user application that needs certain network resources. Different applications have different requirements in terms of bit rate and delay, but also it is important for the network what type of user started that application. Therefore, the message sent from the application that triggers the work of the RM module (StartApp message) has three parameters: userSIM, bitrate and delay. This message is then transferred to the U&AProfile (User&ApplicationProfile). The U&AProfile is the block in which the StartApp message is processed in 5-bit U&A message that identifies the user class (2 bits for 3 possible classes) and the application type (2 bits for 4 application categories) in the Mapping sub block. The last bit in the U&A message is gained from the BatteryCondition sub block and presents additional battery consumption feature (save battery mode and normal battery mode). This bit will be used further to select the mode in which the Decision block will work (one out of three possible). The LRes repository is a database that stashes the information about the targeted networks (newly detected networks). Because the information about possible

1594

networks is available through the MIIS part of MIHF, the LRes has the same structure as the format suggested in the IEEE 802.21 standard. There are around twenty information elements that are defined in the standard, though all of them are not needed for proper functioning of the RM module.

The Recon_Contr is a block left for additional reconfigurability constrains. It can learn and store users’ behavior, thus providing further upgrade of the system (cognition). For example, through this module the decision of the Decision Block can be made much easier - predefined decisions. Also, this block can be used for upgrading users’ profiles. It can send additional messages to the Decision Block (Reconf_rule) to satisfy some users’ preferences. This block will coordinate e.g. exchange of SIP logic based messages to adapt application demands in emergency cases (Adapt_App message). After defining the RIWCoS architecture, its functional blocks and internal communications, the following section will provide a performance analysis of the RM concept within.

Figure 5. RIWCoS’ usage of IEEE 802.21

V.

PERFORMANCE ANALYSIS OF RIWCOS

The ability of the RIWCoS architecture to handle wireless heterogeneity and provision efficient RM is tested on the same simulation scenario as the one on Fig. 1. All RIWCoS introduced functionalities are modeled in a separate executable application communicating with QualNet simulation environment [22]. The RIWCoS approach is compared with the traditional way of serving users, i.e. the SNR based serving policy where mobile nodes camp on the point of attachment having the highest SNR value visible. The performance analysis is conducted by the definition of a novel performance metric, named service retainability, which is able to capture the specific improvements introduced by RIWCoS. This Key Performance Indicator (KPI) parameter is defined as follows. Every mobile node in the simulated scenario requires a certain bit rate from the HWN, R req . The dedicated bit rate for mobile node i when it exhibits a vertical handover j in the simulation scenario is denoted as Ruser ,handover . If the total i

Figure 6. General architecture of RIWCoS’ RM user module

The modes in which the Decision Block can operate are, in priority order: Emergency Mode, BatteryLowMode and Normal Mode. In the Normal Mode, the Decision block does simple switching of the user demands gained from the U&A message to what is available as a resource in the LRes. In the BatteryLowMode (used when the 5th bit in the U&A message is set as ‘1’), the Decision block selects the technology that best fits the battery saving, thus reducing QoS (the selected technology may not be best fitted for the QoS of the application). In the Emergency mode (started when either LinkDown trigger from IM or a custom defined message is received), the RM module uses specially designed algorithm implemented in user and network RM modules for sorting applications and serves them as sorted. The communication with the LRes repository is needed (Get_Info, Push_Info messages) for appropriate design of the ranking list. The Network Discovery module has an interface towards the IM module for receiving MIH messages that carry relevant information about the networks in the users’ vicinity. It uses Store_Info message to fill the LRes database and its work is triggered by the Decision block with the Gather_Info message.

j

number of users is N and the total number of vertical handovers for user i is Mi, then the service retainability is calculated as: N

Service retainability =

Mi

åå

Ruseri ,handoverj

Mi N × Rreq

i =1 j =0

(3)

The goal is to maximize the service retainability as its higher values mean that the mobile node exhibits higher service retention in terms of dedicated bit rate in the HWN. Fig. 7 and 8 present the simulation results for the dependence of the service retainability on the number of mobile users for the RIWCoS and the SNR based serving policies for 64 and 500 kbps application data rates. Fig. 7 shows that the RIWCoS architecture exhibits superior service retainability for 64 kbps application data rates. The gains have values between 20% and 35% depending on the number of mobile users and the users’ mobility. Moreover, it is evident that the service retainability gains are higher for higher mobility making the RIWCoS architecture suitable for military communications. Finally, RIWCoS shows higher service retainability with higher mobility than the SNR based serving policy with lower mobility.

1595

Fig. 8 depicts the RIWCoS architecture behavior in terms of service retainability for 500 kbps application data rates. It is evident that RIWCoS outperforms the SNR based serving policies and that maximum gains are achieved for low number of mobile users (~30%). Comparing Fig. 7 and 8, it is clear that RIWCoS provides performance improvements for both low and high number of mobile nodes (albeit the higher number of mobile nodes exhibits lower performance improvements).

Figure 7. Service retainability comparison for 64 kbps application data rates

triggered when the Decision block in the RIWCoS architecture enters the Emergency operation mode (when an appropriate message from the MIHF is received). SAES is designed to operate in a hybrid mode, i.e. it relies on IEEE 802.21 related information from both the terminal and the network side [10]. IEEE 802.21 messages are used to trigger the algorithm and indicate that the current circumstances require a different way of serving the imperiled users. SAES requires several input parameters such as the number of available interfaces at the terminal side, the number of the active applications in the terminal and the number of exhibited vertical handovers by the terminal prior to the emergency situation in order to do applications prioritization. The current SAES implementation adopts a “fairness” approach that favors the users in emergency which were making lower number of vertical handovers under normal mode of RIWCoS operation. These users are assigned a higher value for a so called priority coefficient [10] that is used in the RIWCoS decision about the serving of the users in emergency (RIWCoS will give priority to users with higher priority coefficient). Extensive details on SAES can be found in [10]. SAES will be analyzed using the same scenario as the one in Fig. 1 without the satellite connection. The nodes for this analysis will be static. The rationale behind is that SAES is triggered once an emergency occurs and the scenario from Fig. 1 is in this case a snapshot of the survived constituents of an HWN. Moreover, there will be two CBR/UDP application sessions per user and the users will be concentrated in the middle of the scenario (where there is a clear overlap of four wireless access networks). The performance analysis will rely on a newly defined KPI metric, service accessibility, able to capture the specific nuances of the SAES algorithm. First of all, the average dedicated bit rate per user is calculated ( Ruser ). If i

the number of sessions is N and the initially requested bit rate per user is Rreq , then the service accessibility is calculated as:

Figure 8. Service retainability comparison for 500 kbps application data rates

i

N

This section clearly showed the benefits of the RIWCoS architecture. Relying on the IEEE 802.21 standard, this architecture is able to cope with RM issues in more efficient manner than the traditional methods proving its soundness for different applications. The specific gains experienced under certain situations, i.e. higher mobility, higher number of users etc., make RIWCoS a potential enabler of more efficient military communications.

Service accessibility =

Ruseri

åR i =1

reqi

(4)

N

Stemming from RIWCoS’ general architecture, the following section will report on a distinct RIWCoS feature allowing efficient coping with emergency/disaster situations which are of great importance for military communications. VI.

EMERGENCY SITUATIONS HANDLING

The previously elaborated RIWCoS operation assumes a normal mode of operation of the Decision block in the RM. A distinct characteristic of the RIWCoS architecture is its ability to efficiently handle emergency/disaster situations, i.e. an abrupt loss of a constituent wireless access network in the HWN. The ability is reflected in the Sorting Applications in Emergency Situations (SAES) algorithm [10] which is

Figure 9. Service accessibility comaprison for 64 kbps application data rates

Fig. 9 and 10 depict the SAES performance compared to the traditional SNR based serving policy under emergency situations for 64 kbps and 640 kbps application data rates, respectively. It is evident that SAES provides higher service accessibility, especially for higher number of sessions. Also, lower application data rates exhibit higher gains.

1596

REFERENCES [1] [2]

[3]

[4] [5] Figure 10. Service accessibility comparison for 640 kbps application data rates

[6]

This section reported on a distinct RIWCoS feature, i.e. the SAES algorithm, and evaluated it in order to prove its feasibility. SAES manages the HWN under emergency in a more intelligent manner providing more resources for the applications. The characteristics of “fairness” can be easily changed by providing different criteria for sessions’ priority coefficients calculation. However, the general conclusions and improvements will still apply.

[7]

[8]

[9]

VII. CONCLUSIONS The emerging of wireless network heterogeneity proves quintessential towards the 4G development. The ability to choose among different wireless access solutions with user transparent and seamless vertical handovers becomes a cornerstone of the HWN increased interest in academia and research. Moreover, the viability of the HWN concept becomes important for military communications, as well as emergency/disaster recovery scenarios. This paper summarized the research efforts on RM architecture for HWNs (RIWCoS). The architecture is based on the IEEE 802.21 standard whose justification was proved through a simulation analysis. The RIWCoS architecture was explained and evaluated (using newly defined metrics) proving its superior performance over traditional users’ serving policies in HWNs. Furthermore, the paper reported on a specific and distinct RIWCoS feature, i.e. the ability to handle emergency/disaster recovery situations through a specially designed algorithm named SAES. Its performances proved to provide higher gains in terms of service accessibility in emergency situations. Future work will include development of network side RIWCoS RM modules, extensive simulation testing of the RIWCoS platform, adding different reconfiguration rules and contexts etc.

[10] [11]

[12]

[13]

[14]

[15]

[16] [17]

[18]

[19]

[20]

ACKNOWLEDGMENT This research is sponsored by NATO’s Public Diplomacy Division in the framework of “Science for Peace” through the SfP-982469 “Reconfigurable Interoperability of Wireless Communications Systems (RIWCoS)” project [5]. The authors would like to thank everyone involved.

[21] [22]

1597

E. Hossain (ed.), Heterogeneous Wireless Access Networks: Architectures and Protocols, Springer, 2009. L. Gavrilovska and V. Atanasovski, “Resource Management in Wireless Heterogeneous Networks,” TELSIKS 2009, Nis, Serbia, October 7-9, 2009. L. Gavrilovska and V. Atanasovski, “Interoperability in Future Wireless Communications Systems: A Roadmap to 4G,” Microwave Review, Vol. 13, No. 1, June 2007, pp.19 - 28. IEEE 802.21: Media Independent Handover. Information available at: http://www.ieee802.org/21. NATO SfP-982469 “Reconfigurable Interoperability of Wireless Communications Systems (RIWCoS)”. Information available at: http://riwcos.comm.pub.ro. L. Gavrilovska and V. Atanasovski, “Reconfigurable Interoperability Towards Seamless Communications,” WPMC2007, Jaipur, India, December 03-06, 2007. O. Ognenoski et al., “Resource Management Aspects in Future Wireless Systems: RIWCoS Approach,” ISABEL 2009, Bratislava, Slovacka, November 24-27, 2009. L. Gavrilovska et al., “Performance of Vertical Handovers in Heterogeneous Wireless Systems utilizing IEEE 802.21,” Workshop “Cognitive Radio Technology and Reconfigurable Wireless Communications”, IEEE Communications 2008 Conference, Bucharest, Romania, June 5-7, 2008. L. Gavrilovska et al., “Providing Interoperability in Heterogeneous Environments towards 4G,” 50th International Symposium ELMAR2008, September 10-13, 2008, Zadar, Croatia. V. Rakovic et al., “A RAT Selection Mechanism in Emergency Scenarios,” Wireless VITAE 2009, Aalborg, Denmark, May 2009. V. Atanasovski and L. Gavrilovska, “Managing Emergency Situations in IEEE 802.21 Enabled Networking Environments,” IEEE EUROCON2009, Saint-Petersburg, Russia, May 2009. G. P. Silvana and H. Schulzrinne, ”SIP and 802.21 for Service Mobility and Pro-active Authentication,” 2008 CNSR, Nova Scotia, Canada, May 2008. T. Buburuzan, “Performance Evaluation of a Handover Model for Integrating Mobile Broadcast Technologies within Heterogeneous Networks,” ISCE '09, Kyoto, Japan, 2009. S.-F. Yang et al., ”A Vertical Media-Independent Handover Decision Algorithm across WiFi and WiMAX Networks,” WOCN08, Indonesia, May 2008. M. Salhani et al., “Terrestrial Wireless Networks and Satellite Systems Convergence,“ 25th AIAA International Communications Satellite Systems Conference (organized by APSCC), AIAA 2007-3288, 2007. C. Fortuna et al., “Demonstration: Wireless Access Network Selection Enabled by Semantic Technologies,” ISWC2009, Washington, 2009. J.Gozalvez et al., “Joint Radio Resource Management in Beyond 3G Heterogeneous Wireless Systems,” WPMC2008, Lapland, Finland, 2008. C.Yang et al., “A Novel Scheme and Access Architecture for Joint Radio Resource Management in Heterogeneous Networks,” IFITA09, China, 2009. Y.Wei et al., “Cooperation Radio Resource Management and Adaptive Vertical Handover in Heterogeneous Wireless Networks,” ICNC08, China, October 2008. M. Alkhawlani and A. Ayesh, “Access Network Selection Based on Fuzzy Logic and Genetic Algorithms,” Advances in Artifical Intelligence, Volume 2008, Article ID 793058, 2008. QualNet Developer. Information available at: http://www.scalablenetworks.com. P. Latkoski et al., “SDL+QualNet: A Novel Simulation Environment for Wireless Heterogeneous Networks,” SIMUTools 2010, Torremolinos, Malaga, Spain, March 2010.