Traffic aware resource allocation with aggregation in heterogeneous networks with wla ns (hlee-2018_eucnc)
0 likes276 views
The document discusses a method for traffic-aware resource allocation in heterogeneous networks, leveraging carrier aggregation to improve Quality of Service (QoS) for various traffic types, including real-time video. A two-step selection process is proposed to allocate resources, enhancing throughput and fairness among users, particularly benefiting those with high mobility. Simulation results indicate significant improvements in system throughput and user satisfaction, highlighting the importance of utilizing both licensed and unlicensed spectrum effectively.
![Overview
Thursday, 5 July 2018 2
q Increase of mobile data traffic
Ø 45% annual growth & ten-fold increase in total traffic by 2021, compared to 2015
<Mobile data traffic by application type (ExaBytes/Month)>
Ø Especially, the growth of real-time traffic (i.e., video) will be very large
[Ref] Ericsson, Ericsson Mobility Report, Nov. 2015](https://image.slidesharecdn.com/traffic-awareresourceallocationwithaggregationinheterogeneousnetworkswithwlanshlee2018eucnc-180705214410/85/Traffic-aware-resource-allocation-with-aggregation-in-heterogeneous-networks-with-wla-ns-hlee-2018_eucnc-2-320.jpg)
![Utility function
Thursday, 5 July 2018 6
q Introduction of utility [14]
Ø It represents the degree of UE’s satisfaction when it acquires a certain
amount of the resource (i.e. meeting the QoS requirements/data rate)
Ø Services with different QoS requirements can be modelled with different
utility functions
q Properties of util. functions
Ø U(0)=0 & U(∞) = 1, U(r) >0
Ø U(r) is an increasing function of
achievable data rate r
Ø Twice differential function with r
Ø U’(r)>0, U’’(r) <0
à U(r) is a concave function
<Example of Utility function for Inelastic traffic>
rmin](https://image.slidesharecdn.com/traffic-awareresourceallocationwithaggregationinheterogeneousnetworkswithwlanshlee2018eucnc-180705214410/85/Traffic-aware-resource-allocation-with-aggregation-in-heterogeneous-networks-with-wla-ns-hlee-2018_eucnc-6-320.jpg)
Traffic aware resource allocation with aggregation in heterogeneous networks with wla ns (hlee-2018_eucnc)
- 1. Traffic-aware Resource allocation with aggregation in Heterogeneous Networks with WLANs Thursday, 5 July 2018 1 Haeyoung Lee, Seiamak Vahid, and Klaus Moessner EuCNC, 18-21 June, 2018 Institute for Communication Systems Faculty of Engineering and Physical Sciences University of Surrey
- 2. Overview Thursday, 5 July 2018 2 q Increase of mobile data traffic Ø 45% annual growth & ten-fold increase in total traffic by 2021, compared to 2015 <Mobile data traffic by application type (ExaBytes/Month)> Ø Especially, the growth of real-time traffic (i.e., video) will be very large [Ref] Ericsson, Ericsson Mobility Report, Nov. 2015
- 3. Introduction Thursday, 5 July 2018 3 q The use of multiple spectrum bands by aggregation Ø Carrier aggregation in LTE-A (up to 32 CCs in Rel-13) Ø Exploitation of unlicensed bands (LAA/LWA) How to utilize radio resource in heterogeneous networks (inc. WLANs) to support the QoS of different traffic types (Inelastic vs. elastic)?
- 4. Ø Multiple carriers in licensed bands are available Freq.F1 F2 F3 … … … Ø Carriers can be utilized through carrier aggregation SC SC PC System model Thursday, 5 July 2018 4 q DL with aggregation for multiple-users in a single cell § carrier aggregation, ü Primary carrier (PC) • Robust connection setup ü Secondary carriers (SCs) • Support of high data rates BS UE 1 UE 2 UE K UE i WiFi AP Ø WLAN links are also available by aggregation. http VoIP video streaming Ø Multiple users requiring heterogeneous traffic types (Inelastic vs. Elastic)
- 5. Objective of proposed approach Thursday, 5 July 2018 5 q To maximize guaranteed QoS while guaranteeing the fairness between UEs SC SC PC BS UE 1 UE 2 UE K UE i WiFi AP http VoIP video streaming Ø Different characteristics of heterogeneous traffic (Elastic vs. Inelastic traffic) Ø Utilization of heterogeneous networks Ø Cellular network having multiple carriers in different licensed bands (PC & SC) Ø WLAN in unlicensed bands (SC only)
- 6. Utility function Thursday, 5 July 2018 6 q Introduction of utility [14] Ø It represents the degree of UE’s satisfaction when it acquires a certain amount of the resource (i.e. meeting the QoS requirements/data rate) Ø Services with different QoS requirements can be modelled with different utility functions q Properties of util. functions Ø U(0)=0 & U(∞) = 1, U(r) >0 Ø U(r) is an increasing function of achievable data rate r Ø Twice differential function with r Ø U’(r)>0, U’’(r) <0 à U(r) is a concave function <Example of Utility function for Inelastic traffic> rmin
- 7. Application Utility Functions Thursday, 5 July 2018 7 q Inelastic traffic Ø Normalized Sigmoidal-like function q Elastic traffic Ø Normalized Logarithmic function <Utility functions for heterogeneous traffics> a, b, k – utility parameters, α – carrier allocation indicator, I – user index, j – carrier index
- 8. Flowchart of prop. approach Thursday, 5 July 2018 8 U1 U2 U3 U4 r2req Allocation of primary carrier in licensed bands {αij, ri pc } Choose UEs requiring more data rates, I = { i | ripc <rireq} Estimate the data rate achievable from WLAN, restWLAN restWLAN => rthWLAN Select the appropriate UEs for WLAN (w/ access index) YES UEs selected for WLAN ? WLAN YES {αij, ri } NO Allocation of secondary carriers in licensed bands NO r4req r1req r3req
- 9. Exploitation of the WLAN Thursday, 5 July 2018 10 q Selection of best suited UEs for WLAN with access index Ø Access index = F(mobility index, signal strength, traffic index) Ø preferred UEs with low mobility, strong WLAN signal, & elastic traffic where is the maxim value of rx. signal power where Ø Calculation of Mobility/Signal/Traffic index
- 10. Simulation Results Thursday, 5 July 2018 11 q Throughput and utility for UEs with different traffic types Ø With different utility function, inelastic UEs tends to be prioritized (two solid lines) Ø With the WLANs access, the system throughput is significantly improved (63.0%) at the high traffic load Ø While elastic traffic is considered more suitable for the WLANs, the QoS of elastic UEs could gain higher than one of inelastic UEs from the use of WLANs (19.5% vs. 6.7%)
- 11. Simulation Results Thursday, 5 July 2018 12 q For different UE’s mobility, Ø The reference algorithm only consider the Wi-Fi signal strength Ø The proposed algorithm gains 39.4% improvement in the average utility and 52% improvement in fairness when UEs move at high speed (avg. 32 km/h) Ø UEs of high mobility tend to be allocated to licensed bands considering limited coverage of WLANs
- 12. Conclusions Thursday, 5 July 2018 13 q Resource allocation in Heterogeneous networks inc. WLANs Ø Utilizes carrier aggregation with two-step allocation considering the characteristics of PC and SCs Ø The WLAN link is used as supplemental carrier as it q Efficient exploitation of unlicensed spectrumof WLANs Ø Need to consider the different characteristics of licensed bands and WLAN links of unlicensed bands. Ø Traffic types, the estimated user mobility speed, and channel quality for each UEs is considered for efficient traffic steering. q Application utility function for Heterogeneous traffic Ø Different utility functions for different traffic types Ø Achieved good QoS even for Inelastic traffics q Further work Ø Need to consider the aspects of different network latency