Coverage and Throughput Analyses of Mobile Multi-hop Relaying Networks P Udhay Prakash & Dr. D Sreenivasa Rao JNTU Hyderabad
[email protected],
[email protected] శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
1
Content for discussion
• • • •
Existing work Single hop vs. multiple hop SCN vs. MCN IEEE 802.16j Networks
– Mutihop cellular networks (MCN) – Mobile multi-hop relaying (MMR) • Relay selection
• Performance Analyses – – – – – –
Success rate Route sustaining time Connection sustaining time Connection duration-outage probability Maximum gain(gmax) Throughput gain
• Concluding where single hop is better and where multihop is better శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
2
Introduction • Current deployments suffer from – Limited Spectrum – Low SINR at Cell edge – Coverage hole due to shadowing
– Non-uniformly distributed traffic load – Unable to address users at cell boundaries, due to power constraints – Allocated 1-2 GHz frequency band is not that suiatble for nLOS communication శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
3
Introduction • Solution: Mobile Multi-hop relaying (MMR) based access network. – Improved data throughput and coverage area with relaying in
cellular
networks.
• Relaying was already used in non-cellular, adhoc networks.
• This paper addresses relaying concept for the cellular networks. • Here, end user can choose to connect directly to a BS, or, a RS, to establish a two-hop link using a relay. • Relay locations are modelled as realizations of a two-dimensional Poisson process with random motion for analyses. శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
4
Existing work •
Mobile radio channel Vary from LoS path to complex path, severely obstructed by buildings, mountains, and foliage.
•
For multi-hop wireless network, a fundamental question is –
to route over many shorter hops (short-hop routing) or over a smaller number of longer hops (long-hop routing).
–
In [4], it is shown that relaying is always not beneficial and the reasons why short hop routing is not as beneficial as it seems to be.
–
In [6], the analysis reveals that multi-hop transmission performs very well in the power-limited regime but can become inefficient in the bandwidth-limited regime without interference cancellation.
–
In [10], the optimal number of hops for a specified end-to-end spectral efficiency (throughput) was analysed for evenly spaced linear networks.
–
In [7, 8], the relative advantages of one hop versus two hop routing were compared, where a deployed relay could provide an improvement in spectral efficiency.
•
In the above literature, –
•
Location of relays is either predetermined or optimized in the design phase.
This presentation focuses on the mobile relays , which was less studied the above literature. శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
5
IEEE 802.16j network •IEEE 802.16 Broadband Wireless Access Working group •IEEE 802.16j supports relay mode operation •Use cases
–Increased coverage
•Extending the coverage range of a BS using multi-hop techniques •Addressing coverage hole problems (e.g., shadows of buildings).
–Capacity enhancement
•Use of multiple links with greater efficiency, as opposed to single-hop links over poor-quality channels. •Multi-hop communications, which can support spatial reuse [9].
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
6
SCN vs. MCN •
Hop: The step from one router to the next, on the path of a packet on any communications network. SCN: Single-hop Cellular Network
•
MCN:Multihop cellular network
•
–
Infrastructure-based cellular networks with adhoc networking concept
–
SCN++
–
Fixed Base Stations + Adhoc networking
–
Enhanced coverage, improved capacity and flexibility.
–
Mobile relays are not (yet) of practical interest except in some specific applications such as professional radios for emergency response, police and security organizations.
–
Provides cellular systems with opportunity of peer-to-peer (mobile to mobile) communication as well as communication relayed through other fixed and/or mobile terminals.
•
The increase in system throughput is the major advantage of MCN.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
7
SCN vs. MCN •
•
SCN –
BSs must be reached by MSs in a single-hop.
–
Subcell in SCN -- area of a sub cell is the same as the area of a cell.
MCN –
Cell radius is half the distance between two neighbouring BSs.
–
BSs need not always be reachable by MSs in a single hop.
–
sub cell in MCN -- area reachable in a single wireless hop by a BS or a MS
–
BS and MSs are not always reciprocally accessible in a single hop.
–
transmission range of BS and MSs can be reduced than that in SCNs.
–
accessible area by a BS or a MS is the area of a sub-cell.
–
MSs can directly communicate with each other provided that they are mutually reachable and belonging to the same cell.
–
perform multi-hop routing.
–
when destination MS is in a different cell from that of the source MS, then the Relay Station forwards the packets to its own BS, which in turn, forwards to the destined MS via its BS.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
8
MMR •
MMR-Mobile multihop relaying
•
Concept of relaying user data and possibly control information between an MMR-BS and MS through one or more relay stations (RS).
•
Mobile—because both RS & MS are mobile.
•
Relaying – To enhance coverage, range, and throughput and possibly capacity of an MMR-BS – To enable very low power devices to participate in the network.
•
Multipath routing between the MMR-BS and an MS to communicate user data and/or control/management information, to improve communications reliability.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
9
Relay selection •
Relay Selection effects hop delay and the complexity involved.
•
Assumed variables are –
dij(t)—distance between mobile i and mobile j at time t. Index 0 denotes the BS.
– di0(t) & dj0(t)—distances between mobile i and BS, & mobile j and BS respectively at time instant t. – r –transmission range for any mobile. – Clearly, d00(t)=0 for all t.
– M= { 1,2,..N }—set of mobiles and N= { 0,1,2,…, N}—set of nodes including the BS (i=0). – R(t)ϵN—set of relay nodes at time slot t. – A(t)ϵ M—set of active nodes. i.e, the nodes that are not acting as relays.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
10
Relay selection • Node i ϵ A(t) selects relay ki as Ki = argminjFi(t){λdij + (1-λ )dj0} for all iϵ A(t) • where 0≤ λ≤ 1 is a weighting parameter • Fi(t)—set of feasible relays for mobile i. • Fi(t) ={ j/j ϵR(t); dij< r; dj0≤ di0}
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
11
Performance Analyses - Simulation consideration
• • •
Simulations were performed in MATLAB 7.12. End user is fixed at coordinates (l,0) and Mobile relay is at coordinates (r,θ) Since BS coverage has been normalized, – –
• • • • • • • •
M/M/∞ queuing model is used to capture relay mobility L –distance between BS and end user r—distance between BS and mobile relay θ- angle between BS and mobile relay with end user λ1and λ2—SNRs of BS and mobile relay respectively α –path loss exponent N—average no. of usable relays in cell coverage area N=πρ, with relay density ρ – –
•
l >1 corresponds to out-of-coverage users, l ≤ 1 implies the end user is within the coverage area.
Assuming N = 20 to represent a low density cell N = 100 to represent a high density cell for numerical evaluation.
Average relay speed is normalized to the cell diameter శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
12
Performance Analyses
Simulation Environment with respective Nodes’ coordinate positions. • 20 nodes (node may be MS, BS or a relay) forming a network, with every nodes connected to its nearby nodes, is created. • The relay movement is randomized in distance and direction. శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
13
Success rate •
For r = 10, θ = 35, and SNR of BS and
relay is λ1= λ2= 3dB with α= 3, •
Simulations for l = 1.05, 1.10, 1.15 and 1.20, for out of coverage end user.
•
As
feasible
region
shrinks
with
increasing l, chance of locating a relay within the feasible region also declines.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
14
Success rate • Impact of SNRs λ1 and λ2 on two-hop routing success probability is depicted here for an end user 10% away from the BS coverage area l = 1.1 and considering λ1= 2 dB, 3 dB, 4 dB and 5 dB. • For two hop relaying to be useful, the relay SNR at unit distance λ2 should be closer to that of the BS, λ1, as λ1 increases. • Even with λ1= λ2, higher SNRs reduce probability of feasible relays. • Two-hop relaying is less favourable in high SNR regions for line networks.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
15
Route sustaining time • Once the route (path) is established, its sustainability is analysed, either with or without re-routing. • Assuming that handoffs between relays are allowed and they must be in time. • Here l = 1, 1.05, 1.1, 1.15, and 1.2, with pedestrian speed as 2 and vehicular speed as 10, with α= 3 and λ1= λ2= 3 dB. • Average route sustaining time is much longer than average burst duration in IEEE 802.16j network architecture, even when the mobile relay travels at a vehicular speed.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
16
Connection sustaining time • •
•
For user to BS distances l =1.1 Tn —connection sustaining time, averaged over all time instants for n feasible relays. route sustaining time >> connection sustaining time
– due to possibility of new relays entering the feasible region as current feasible relays leave. – So, allowing mobile relay hand-off is an effective method to extend connection time.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
17
Connection sustaining time • For user to BS distances l =1.2 • For vehicular speed and pedestrian speed users • When relay density is high, average connection sustaining time for l = 1.1 is very long – on the order of days, depending on maximum time.
• connection success rate approaching one as the relay density increases.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
18
Connection duration-outage probability •
Probability that two-hop connection fails to meet connection duration requirement due to depletion of feasible mobile relays.
•
Exponentially distributed with mean given by the x-axis value,
•
Mobile relays are distributed with
poisson
point process such that there are an average N = 20 relays in the cell. •
Mobile relays are assumed to move at
pedestrian speed, with α= 3.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
19
Maximum gain(gmax) •
• • •
• • •
For α = 4 with λ1= 3 dB, λ2= 1 dB, 2 dB and 3 dB and for α= 3 with λ1= 3 dB, λ2= 1 dB, 2 dB and 3 dB. Gmax can be determined by searching for the optimum relay position. Graph depicts how Gmax varies with user distance l. Assumin α has a big impact on Gmax. Gmax increases with increasing l, as end users close to BS already enjoy a high throughput. So, multi-hop is less favourable in the high spectral efficiency regime.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
20
Throughput gain • For α = 3 , and end user is located at cell boundary. • With the upper bound Gmax,
simulation results show how random relay placement affects the throughput gain.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
21
Throughput gain • For α = 4, and end user is located at
cell boundary. • With
increasing
relay
density,
probability that relaying achieves a
gain close to Gmax increases. • And, the average throughput gain approaches Gmax.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
22
Conclusions •
Concluding where single hop is better and where multihop is better.
•
For the two-hop links, – success rate is inversely proportional to the coverage distance. – two hop networks are unfavourable in high SNR regions for line networks. – Average route sustaining time is much longer than the average burst duration. – Connection sustaining time is directly proportional to the relay density and inversely proportional to the relay speed. – With increasing relay density, the achievable throughput reaches the maximum gain level.
•
For an out of-coverage end-user, mobile relays offer substantial coverage extension benefits.
•
With randomly placed moderate number of mobile relays, significant average throughput gains can be obtained for end users near cell boundaries.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
23
Future Scope • Above presented work can be extended for different SNR values. • Power consumption and security aspects of relay supported cellular networks can be analysed. • Alternative techniques for delay reduction such as decreasing packet size can be analysed. శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
24
References 1)
Y. D. Lin and Y. C. Hsu, Multi-hop cellular: A new architecture for wireless communication, in IEEE INFOCOM '2000, 2000, pp. 12731282.
2)
Andrea Goldsmith, Wireless Communications, Cambridge University Press, 2005.
3)
Halim Yanikomeroglu, Cellular Multi-hop Communications: infrastructure Based Relay Network Architecture for 4G Systems, Queen's University 22nd Biennial Symposium on Communications (QBSC'04), Kingston, Ontario, Canada, 1st June 2004.
4)
M.
Haenggi and D.
Puccinelli, Routing in adhoc networks: a case for long hops, IEEE
Commun. Mag., vol. 43, pp. 93-101, Oct. 2005. 5)
A. Florea and H. Yanikomeroglu, On the Optimal number of hops in infrastructure-based fixed
relay networks, in Proc. IEEE Globecom, St. Louis, MO, Nov. 2005.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
25
References 6)
M. Sikora, J.N. Laneman, M. Haenggi, D.J. Costello Jr., and T. E. Fuja, Bandwidth and Power efficient routing in linear wireless networks, IEEE Transactions on Inf. Theory, Vol.52, pp. 26242633, June 2006.
7)
S. V. Maiya, Spectral efficiency and its relation torouting strategies in simple communication networks, Master's thesis, University of Notre Dame, Notre Dame, 2007.
8)
S.V. Maiya and T.E. Fuja, One hop vs two hop routing in simple networks with fading: an outage probability analysis addressing spectral efficiency, in Proc. Wireless Communication Networks Conference (WCNC 2008), Las Vegas, Mar. 2008.
9)
Vaskengenc, Seanmurphy, Yangju and Johnmurphy, IEEE 802.16j Relay-based Wireless Access Networks: An Overview, IEEE Wireless Communications, October 2008.
శనివారం, 10 ఆగస్టు 2013
ICIECE-13 ఉదయ్ పకాష్
26