Context-aware Middleware forMulti-hop Multi-path
Heterogeneous Connectivityin Social Sharing Scenarios
18.09.2009DEIS, Università degli Studi di Bologna,
Viale Risorgimento, 2 - 40136 Bologna [email protected]
Carlo Giannelli
Bologna, Italy — 18.09.2009 Carlo Giannelli 1/41
AgendaSocial sharing of connectivity resources
– single/multi-hop wireless networks– sharing of Internet connectivity and peer-to-peer services
State-of-the-art thorough analysis– CAMPO model and taxonomy– from traditional homogeneous to novel heterogeneous wireless scenarios
several communication technologiesinfrastructure and peer points of access/services
Multi-hop Multi-path Heterogeneous Connectivity (MMHC)– middleware for context-aware dynamic reliable connections to the Internet– context information: node mobility, path throughput, energy availability– MMHC vs. IEEE 802.11s– two-phase procedure
local-phase: reliable remote connection establishmentglobal-phase: available paths enhancing to ensure long-term availability
– MMHC middleware architecture
Ongoing work– from Internet-based to p2p-driven networking– incentives to resources exploitation fairness in semi-cooperative environments– cluster-based mobility and smart environment management
Bologna, Italy — 18.09.2009 Carlo Giannelli 2/41
The Wireless ScenarioClient: node requiring connectivity, e.g., user PDAConnector: node providing connectivity, e.g., UMTS Base Station (BS)Channel: active client-connector IP connection, e.g., IEEE 802.11 association and DHCP configuration
Handover procedure– a client node changes current connector while
movingEvaluation process
– context gathering: which information is important?
– metric application: which is the most suitable connector?
Internet
Connectors
Channels
ClientNode
Bologna, Italy — 18.09.2009 Carlo Giannelli
Single/Multi-hop Wireless Networking
Single-hop networking: direct communication among nodes– ad-hoc: point-to-point communication
Bluetooth Piconet or IEEE 802.11 IBSS– infrastructure: communication mediated by special purpose
equipment IEEE 802.11 AP or UMTS BS
Multi-hop networking: communication among distant nodes based on packet routing performed by intermediate nodes– different hypothesis on
availability of infrastructure componentsmobility degree of communicating and intermediate nodes
3/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
WMN, MANET, Opportunistic Networking
4/41
Wireless Mesh Network, WMN– intermediate nodes are mainly infrastructure-based, aiming at creating highly reliable
backbones, e.g., IEEE 802.11s– communicating nodes move frequently, intermediate nodes are rather static
Mobile Ad-hoc NETwork, MANET– intermediate nodes are based on ad-hoc connectivity: best-effort connectivity– both communicating and intermediate nodes move frequently, but there is enough time
to create sufficiently reliable paths among nodes
Opportunistic Networking– similar to MANET, but without a path among communicating nodes– intermediate nodes opportunistically forward packets whenever interact with other
nodes “closer” to the destination: suitable in unreliable environments
backbone
Bologna, Italy — 18.09.2009 Carlo Giannelli
Social Sharing of Connectivity Resources (1)
5/41
Novel scenario taking full advantage of the many context information and networking opportunities
peer-to-peer connectivitysimultaneous exploitation of multiple interfacesaverage node mobility (between MANET and WMN)
Internet sharingInternet connectivity provisioning based on dynamic context-aware routing rules configuration
Peer-to-peer service sharingtime/hop-bounded service discovery/invocation in heterogeneous ad-hoc networks
Quality of Service controlincentive-based to push for actual connectivity/service sharing
Bologna, Italy — 18.09.2009 Carlo Giannelli
Social Sharing of Connectivity Resources (2)
6/41
Bluetooth
Internet
UMTSIEEE 802.11
IEEE 802.11(ad hoc)
Carol
Alice
Alice
Bob
(Alice moves)
AP BS
1) Alice’s client autonomously selects the free-of-charge IEEE 802.11 connector
2) Alice provides connectivity via Bluetooth
3) Carol provides her lesson’s notes via a file sharing service
4) Bob accesses Carol’s notes via Alice
5) When Alice moves, she gets connectivity via Carol and then re-establish active connections
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Model (1)Thorough state-of-the-art surveyMany specific research areas, e.g., infrastructure-based and peer-to-peerNovel model and taxonomy:
– proposals grouping based on differences and similarities
– novel researchers field comprehension
CAMPO: Context-aware Autonomic Management of Preferred network Opportunity
– most suitable interface and connector based on user preferences, runtime environment, connectivity reliability
A common model for– basic term definition
channel: {application, interface, connector}intra/inter/vertical handover
– two selection mechanismsinterface and connector selector
Application
Active InterfaceNetworkDomainUnused Interface
Connector
ba
cc
Applications ConnectorsInterfaces
a) Intra-horizontal, b) Inter-horizontal and c) vertical handover
7/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Model (2):4G and ABC
4G (4th Generation)– one interface per time, selected without
user intervention <N:1:1>– infrastructure-based connectors– dynamic connection migration among
interfaces
ABC (Always Best Connected)– several interfaces simultaneously– infrastructure and peer connectors– only-one/multiple channels per interface
<N:M:M> / <N:M:L>
Social Sharing– Internet connectivity + peer-to-peer services– user's behavior active monitoring– reward-based mechanisms
8/41
N 1 1 1
cc
<N:M
:M>
<N:M
:L>
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Taxonomy (1)Three categories to underline specific similarities and differences:– management scope
working environment capabilities– evaluation process
context gathering: which information are important?metric application: which interface/connector is the most suitable one?
– continuity management trigger: when performing a handover?switcher: how to update active channels?
9/41
interfacemobile node
environment
Internet
contextgathering
metricapplication trigger switcher
Evaluation Process Continuity Management
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Taxonomy (2)
Management scope:– interface: switch on/off, select a connector– mobile node: only one/multiple interfaces– environment: external components support
10/41
distributed
single-on
envi-ronment
interface
client-side
interface-only
interface-connector
infrastructure
peer
location
roleevaluation
continuity
mobilenode multiple-on
ManagementScope
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Taxonomy (3)Primary operations of CAMPO systemsEvaluation process– available channels suitability
input: context informationprocessing: how to exploit inputoutput: channel suitability, best channels
Continuity management– update active channels at service
provisioning timeintegration: relationship between origin and destination connectorsgranularity: every-/per-channel migrationvisibility: external support
11/41
ContinuityManagement
per channel
per node re-routing
endpoint updategranularity
visibilityend-to-end
transparent
integrationtight
loose
proxy
intra-domaininter-domain
intra-horizontalinter-horizontal
vertical
pureproxy
re-addressing
AAA, billing
extensible
abstractionlevel
flexibility
variations
objective
dynamic
embedded
physicalnetwork
application
selectedentity
interfaceconnector
localglobal
static
input
processing
output provideddata
single valueevaluation set
EvaluationProcess
Bologna, Italy — 18.09.2009 Carlo Giannelli
CAMPO Taxonomy (4)
A survey positioning about 80 work– many systems provide only partial solutions– deployment scenario: connector scope, single-on– evaluation process: local scope, function, connector– continuity management: loosely-coupled, per-node, proxy
12/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Lessons Learned
Context-aware evaluation process– dynamic adaptation of the execution environment– trade-off between effectiveness and expressiveness
Hybrid deployment scenarios– infrastructure and peer connectors– multi-hop multi-path heterogeneous connectivity
Decentralized connectivity management– collaborating components distributed on different
nodes, eventually even on the infrastructure side
13/41
Bologna, Italy — 18.09.2009 Carlo Giannelli 14/41
Homogeneous Wireless Scenario
Current scenario: static, infrastructure-based, one-hopOne communication interface at a time
– the client node does not change wireless interfaceHorizontal handover
– infrastructure connectors only– origin and destination connectors based on the same wireless technology
IEEE 802.11– connectors are IEEE 802.11 Access Points (APs)– metric based on Received Signal Strength Indication (RSSI) and Signal
to Noise Ratio (SNR), usually embedded in interface firmware
Already available many heterogeneous wireless interfaces on the same mobile node– IEEE 802.11, Bluetooth, GPRS/UMTS– bandwidth, power consumption, coverage range
Bologna, Italy — 18.09.2009 Carlo Giannelli 15/41
Social Heterogeneous Wireless Scenario
Heterogeneous interfaces– the client node exploits multiple wireless
interfaces, even simultaneouslyHeterogeneous connectors
– can be infrastructure or peer nodes– single-/multi-hop paths
Connectivity management– managing interfaces/connectors/channels/paths
considering several context data to take advantage of the many networking opportunities
Heterogeneity and cooperation increases client node capabilities:– heterogeneous connectors enable the most suitable form of connectivity
Bluetooth to limit power consumption, IEEE 802.11 to get larger bandwidth– peer connectors extend connectivity opportunities via multi-hop paths
UMTS link accessed via Bluetooth through a peer connector
peer
infrastructure
Internet
IEEE 802.11
Bluetooth
UMTS
Client Node
peerconnector
Bologna, Italy — 18.09.2009 Carlo Giannelli
Social Scenario:Design Guidelines
Trade-off between static and dynamic management– re-evaluation of connectivity opportunities to
modify available channelsreconfigure routing rules
– static approaches achieve sub-optimal solutions– dynamic approaches impose non-negligible monitoring/managing overhead
Trade-off between local and global management– monitoring scope greatly impact on required solution
local knowledge: easy to gather but with limited expressivenessglobal knowledge: best resource exploitation but with networking overhead and delay
Trade-off between single- and multi-path granularity– basic solution: every flow of every client sent to the same destination
aggregated routing rules – client-granularity: different flows of the same client sent to the same destination
per-client routing rules– flow-granularity: each flow is managed differently
per-request routing rules16/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC: Multi-hop Multi-pathHeterogeneous Connectivity
Full exploitation of already available wireless interfaces– multi-hop paths, eventually based on heterogeneous single-hop links– dynamic connectivity evaluation and management– connectivity provisioning in a peer-to-peer fashion– context-aware evaluation metric
Efficient connectivity support via proper trade-off among – static and dynamic management
time-consuming single-hop connections performed reactively (rather static)efficient routing rules modifications performed proactively (dynamic)
– local and global managementlocal management to quickly provide Internet connectivityglobal management to incrementally improve connectivity capabilities
– single- and multi-path granularityaggregated connectivity to the Internetdifferentiated access to peer to peer services
17/41
Bologna, Italy — 18.09.2009 Carlo Giannelli 18/41
MMHC: Objectives1) Novel metric considering a wide set of context information at
different abstraction levelstraditional RSSI/SNR based evaluation processes are not enoughevaluation metric specifically designed for heterogeneous wireless scenarios
2) Two-phase procedure to separately consider path establishment and enhancement
local-phase: connectors suitable for path realization to maximize reliability and throughputglobal-phase: long-term connectivity based on additional context information, eventually slight modifications of the network topology
3) Static and dynamic managementreactive approach for single-hop connectivityproactive approach for multi-hop path reconfiguration
Bologna, Italy — 18.09.2009 Carlo Giannelli
IEEE 802.11s for Extended Service Set mesh networking
Multi-hop connectivity at lower OSI layers, by extending IEEE 802.11
– efficiency: no en/decapsulation of data into/from higher layer protocols
– availability: compatibility with the many already available IEEE 802.11a/b/g interfaces
Node roles– STA: mobile client STAtion only getting connectivity– MAP: Mash Access Point providing connectivity to STAs– MP: Mesh Point performing as intermediate node (not
providing connectivity to STA)– MPP: Mesh Portal interconnecting the mesh network and
the InternetPath establishment
– metric: airtime cost reflecting the amount of channel resources consumed by transmitting a frame over a particular link
– path selection: hybrid reactive/proactive
19/18
Internet
MPP
MPMAP
MAP
STA
STA STA
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC vs. IEEE 802.11sBoth IEEE 802.11s and MMHC support multi-hop wireless connectivity, but with relevant differencesRoles
– IEEE 802.11s: STA, MP, MAP, MPP– MMHC: clients, (peer) connectors
Technology– IEEE 802.11s: only standard-compliant interfaces– MMHC: interconnection of heterogeneous networks
Layer– MAC protocol– interconnection of IP networks
Evaluation metric– IEEE 802.11s: low-level radio-aware link metric– MMHC: meaningful context information
IEEE 802.11s and MMHC are complementary, not competitors
20/41
Internet
MPP
MPMAP
MAP
STA
STA STA
premere sulla necessità di
sfruttare il contesto
Bologna, Italy — 18.09.2009 Carlo Giannelli
Context AwarenessProvide highly reliable/durable paths (crucial issue in mobile wireless networks) with sufficient quality (to maximize user satisfaction)Path reliability– peer connectors are less reliable, since may abruptly move away– monitor client node and peer mobility to provide reliability
Path quality– quality mainly depends on wireless technology, number of
active clients, and number of hops to the Internet – coarse-grained estimation of actual throughput
Path durability– interrupt the connectivity to limit power consumption– residual battery level to ensure path long-term durability
21/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
ConnectorsInfrastructure-based connectors– e.g., IEEE 802.11 access point and UMTS base station– always reliable and fixed
Peer-based connectors– e.g., IEEE 802.11 ad-hoc and Bluetooth– trusted and untrusted– fixed and mobile– joint and transient
connector
jointtransient
fixed
peer(un/trusted)
mobile
infrastructure(trusted)
22/41
Bologna, Italy — 18.09.2009 Carlo Giannelli 23/41
Transient connector– e.g., a mobile node in the same sidewalk but with opposite direction– not suitable for connectivity since has a high probability of becoming
unavailableJoint connector
– e.g., PDA connector in the same train wagon– greater durability → suitable for connectivity
Client-connector mutual distance inferred by monitoring connector RSSI variability
– CMob to evaluate client node mobility degree [0,1]– Joint to evaluate peer connector relative mobility degree [0,1]
ClientNode
joint
transient
Path Reliability (1)
Connector type RSSI variability Mobility state
fixedalmost constant still client nodegreatly variable moving client node
mobilealmost constant joint connectorgreatly variable transient connector
Bologna, Italy — 18.09.2009 Carlo Giannelli 24/41
RSSIgathering
Low-PassFilter
ParameterComputation
StateEstimation
DFT (16/4 values)IDFT (first harmonic)
DFT(16/4 values)
RSSIsequences
filtered RSSI
sequences
firstharmonic modules
CMobJoint
upper/lowerbounds
Discrete Fourier Transform (DFT) applied twice to– low pass filter RSSI fluctuations due to signal noise– estimate CMob (fixed infrastructure connectors) and Joint (peer
connectors)
Path Reliability (2)
Single-hop: EstimatedEndurance– (1-CMob) • CoverageRange (for APs/BSs)– Joint • CoverageRange (for mobile peers)
Multi-hop: PathMobility at kth hop– EstimatedEndurancek (single-hop, i.e., k=1)– EstimatedEndurancek • PathMobilityk-1 (multi-hop, i.e., k>1)
Bologna, Italy — 18.09.2009 Carlo Giannelli
Path Reliability (3)
Node B and C are still → high EE
Node A is in motion → low EE
Node D automaticallyselects the path on the right
– nodes B and C have same EE but different PM
– node C provides much higher reliability (PMB=0.49 vs. PMA=0.14)
25/41
A
D
InternetInternet
AP1 AP2
C PMAP2 = 0.7
EEAP2 = 0.7EEAP1 = 0.7
PMAP2 = 0.7
EEA = 0.2
EEB = 0.7
PMC = 0.49
PMA = 0.14
Lessons learned: push for paths composed by joint nodes
B
PMB ≈ 0.01
EEC = 0.7
Bologna, Italy — 18.09.2009 Carlo Giannelli
Path Quality (1)Coarse-grained estimation of multi-hop paths throughput
– adopted wireless technology: e.g., Bluetooth represents a bottleneck– number of active clients: fair bandwidth sharing– number of hops to the Internet: 20-30% per-hop degradation
Heterogeneous wireless interfaces provided by different manufactures, e.g., IEEE 802.11 Orinoco Gold, Buffalo and PRO/Wireless interfaces
– heterogeneous interfaces better mimics actual wireless environments– greater performance with homogeneous hardware
26/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Path Quality (2)
EstimatedThroughput (ET):– NominalBandwidth (NB) (for APs/BSs)– (1 – HopDegr) • MaxThr / #clients (for mobile peers)
where MaxThr=min{previous hop ET, current hop NB}
– ETAP = NBAP = 4 Mbps– ETA = (1-0.2) • 4 Mbps / 3 clients = 1.07 Mbps– ETB = (1-0.2) • 1.07 Mbps / 2 clients = 0.428 Mbps
Lessons learned: push for short paths with few clients, particularly when exploiting Bluetooth
InternetAP
A
B
27/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Path Durability (1)Expected long-term path durability due to energy consumption
– avoid paths composed by mobile peers with low battery levelsprobably unavailable in a short time
– fairly exploit energy of mobile peers not overloading only one path traversing traffic increase power consumption
ResidualPathEnergy at kth hop– NodeBatteryLevelk (single-hop, i.e., k=1)– NodeBatteryLevelk • ResidualPathEnergyk-1 (multi-hop, i.e., k>1)
AveragePathEnergy at kth hop– NodeBatteryLevelk (single-hop, i.e., k=1)
– (multi-hop, i.e., k>1)1( ) ( 1)k kAveragePathEnergy k NodeBatteryLevelk
− ⋅ − +
28/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Path Durability (2)
A
C D
E
InternetInternet
BS1 BS2
F
B NBL = 0.40
APE = 0.45RPE = 0.20
NBL = 0.07
NBL = 0.95
APE = 0.40RPE = 0.40
NBL = 0.50
APE = 0.95RPE = 0.95
APE = 0.51RPE ≈ 0.07
NBL: NodeBatteryLevelRPE: ResidualPathEnergyAPE: AveragePathEnergy
F selects BS2-B-D instead of BS1-A-C
– slightly lower APE: 0.45 instead of 0.51
– but sufficiently great RPE: 0.20 instead of 0.07
Lessons learned: push for battery level fair exploitation
29/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Ongoing WorkCurrent MMHC prototype is mainly focused on Internet connectivity
– main goal is to provide multi-hop Internet access to nodes– peer-to-peer communication limited to subnets
Extending Internet connectivity provisioning via peer-to-peer networking
– incremental knowledge of the (whole) wireless environment– local (vs. Internet) service provisioning/discovery/invocation
Pushing for service/connectivity provisioning– supporting trust and fairness in semi-cooperative environments
30/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Peer-to-peer Information Delivery
Supporting context information spreading among NAT-separated but interconnected networks
– not only top-down but also bottom-up and sibling information delivery
Information delivery with no global knowledge of network topology– multiple single-hop information dissemination among
neighbor nodes– time/hop bounded information delivery
Service discovery/provisioning as a special case of context distribution– context-aware service discovery/selection– automatic service rebinding
31/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Connectivity Fairness in Multi-hop Wireless Networks
Connectivity starvation in multi-hop multi-client paths
– closest node achieves almost all the bandwidth
Decentralized fairness management– traffic monitor to perceive starvation– traffic control to
maximize local trafficavoid traffic starvation
Incentives for connectivity sharing based on low/high level context information
– traversing vs. generated packets– number of invoked/offered services– number of connected nodes
32/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Conclusions & Ongoing Work
MMHC supports multi-hop multi-path connectivity exploiting off-the-shelf heterogeneous equipment– IEEE 802.11, Bluetooth, Ethernet
MMHC proposes innovative context data suitable for heterogeneous wireless scenarios
– node mobility, path throughput, energy availabilityMMHC main goal is to provide reliable connections in wireless mobile environments– throughput as secondary objective
Ongoing work– from Internet-based to peer-to-peer connectivity– security issues: peer mutual authentication, user incentives,
dynamic level of trust management– spontaneous smart environments based on dynamic clustering
33/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Any question?
Prototype code and implementation insights:– http://lia.deis.unibo.it/research/MAC/– http://lia.deis.unibo.it/research/MACHINE/– http://lia.deis.unibo.it/research/MMHC/– http://lia.deis.unibo.it/Staff/CarloGiannelli/
Bologna, Italy — 18.09.2009 Carlo Giannelli 35/18
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC Local PhaseMain goal: quickly achieve connectivity to the Internet
– locally gathers RSSI and estimates CMob/Joint– performs single-hop reliable connections based on
EstimatedEndurance (completely distributed evaluation)– select the most suitable path based on PathMobility and
EstimatedThroughput (distribution of few crucial context information)
Local phase path selection metric: select the path with best trade-off among PathMobility and EstimatedThroughput:
– avoid highly unreliable paths– maximize throughput
Reactively activated at path disruption
Greater priority to connection reliability than quality
36/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC Global PhaseMain goal: ensure long-term availability enhancing connectivity capabilities
– periodically interact with nearby node to collect PathMobility, EstimatedThroughput, AveragePathEnergy and ResidualPathEnergy
– trigger path modification whenever a link is broken (reactive), ResidualPathEnergy lowers below 0.1 (proactive), or PathMobilitylowers below 0.3 (proactive)
– select the path with best trade-off among EstimatedThroughput and AverageBatteryEnergy
avoid nodes with low battery levelfairly exploit available pathsachieve high connectivity quality
Maximize user perceived quality of service– available paths periodic monitoring and proactive reconfiguration– enhance connectivity opportunities via the role-switch procedure
37/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Role-switch ProcedureA client can work as bridge among different networks
– the peer connector contribute providing the physical network, e.g., performing as Bluetooth master
– a client starts forwarding data via one of its available paths: it acts as a gateway
Note: role-switch aim at providing only Internet connectivity
F has two paths to the InternetWhen A fails C exploits F as gatewayBoth C and E keep connectivity to the Internet via F
C D
E
InternetInternet
BS1 BS2
F
BA
38/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC Design
Full NetworkExploitation
SuitablePaths
SuitableChannels
MobilityDegree
Connector monitoring and interface managing(mobile client requirements, e.g., user and OS)
Channel monitoring and routing rule managing(mobile client and remote node requirements)
peer-to-peer service provisioning
Met
ric A
pplic
atio
n
Raw data gathering andmobility degree estimation
RoutingManager
ConnectorManager
P2PManager
Mobility& Peer
Estimator
ComponentName
ProvidedFeature Performed actions
Eva
luat
ion
Pro
cess
AvailableConnectors
Low-level components uniform and aggregated access provisioning
NetworkInterfaceProviderC
onte
xt G
athe
ring
BestPath
One-shot selection among available channels(application specific requirements)
PathApplication
Selector
39/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC ArchitectureNetwork Interface Provider– homogeneous access to
heterogeneous interfaces on different operating systems
Connector Manager– single-hop connections based
on node mobility
Routing Manager– context information remote
distribution– multi-hop paths based on
estimated connectivity availability and throughput
Network Interface Providerpr
ovid
e co
nnec
tivity
single-hopopportunities
multi-hoppaths
ConnectorManager
connect
IEEE 802.11 Bluetooth UMTS
RoutingManager
local noderequirements
single-hopconnections
receive remote context
send local context
40/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Network Interface ProviderNetwork Interface Provider (NIP) provides a homogeneous access to heterogeneous interfacesFeatures: set of capabilities common to interfaces
– get available connectors, to get available connectors list and related information such as RSSI
– perform as peer connector, to offer connectivity in a peer-to-peer fashion
– connect to a connector, to perform a connection with a given connector
Network Interface Provider
provideconnectivity
single-hopopportunities
connect
IEEE 802.11 Bluetooth
IEEE 802.11 standard– scan of available ESSID (Extended
Service Set ID)– connectivity provisioning via IBSS
(Independent Basic Service Set)– association to a given AP
41/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Connector ManagerConnector Manager (CM) establishes single-hop channels with remote devices
1) connectors discovery via any interface2) connectors evaluation based on mobility degree3) requires layer2 connections with most suitable connector of
each interface4) activates layer3 configuration via DHCP client
CM does not interact with remote nodes– mobility degree achieved locally in
a lightweight manner– requirements in terms of maximum
node-connector mutual mobilitysingle-hopopportunities
ConnectorManager
connect
local noderequirements
single-hopconnections
42/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
Routing ManagerRouting Manager (RM) handles multi-hop paths to the Internet1) context information exchange with one-hop distant nodes2) single-hop distant links evaluation3) routing rule modification to provide suitable multi-hop
paths
multi-hoppaths
RoutingManager
single-hopconnections
receive remote context
send local context
RM has a wider perspective– mobility, throughput and energy
of paths– discard unreliable paths due to
mobility, then achieve a trade-offamong throughput and energy
43/41
Bologna, Italy — 18.09.2009 Carlo Giannelli
MMHC Performance ResultsTime consuming single-hop creation and efficient path reconfiguration
– CM: reactive (rather static) – RM: proactive (greatly dynamic)
MMHC overhead is negligible– connection establishment delay mainly
due to specific characteristics of wireless technologies
0200400600800
1000
new path path recovery
path change
Res
pons
iven
ess
(mill
isec
onds
)
Routing Manager
3,04
14,37
0,14
0,12
0,12
3,43
2,04
3,29
0,00
2,00
4,00
6,00
8,00
10,00
12,00
14,00
16,00
18,00
20,00
22,00
IEEE 802.11 Bluetooth
Cum
ulat
ive
Tim
e (s
econ
ds)
Automatic DHCP configuration
Wi-Fi association/ Bluetooth PAN connection MMHC connector evaluation
Wi-Fi scan/ Bluetooth inquiry
Connector Manager44/41