Overview • Fundamental issues and impact • wireless • mobility • For each layer in the protocol stack • A subset of design requirements • Design challenges/constraints • Possible design options
F E B C A D Physical Properties of Wireless • Wireless = Waves, electromagnetic radiation emitted by sinusoidal current running through a transmitting antenna • Signal will be received by everyone nearby. • Makes wireless network different from wired networks
Typical Radio System(Sender) • A radio system transmits information to the transmitter. • The information is transmitted through an antenna which converts the RF signal into an electromagnetic wave. • The transmission medium for electromagnetic wave propagation is free space.
Typical Radio System(Receiver) • The electromagnetic wave is intercepted by the receiving antenna which converts it back to an RFsignal. • Ideally, this RF signal is the same as that originally generated by the transmitter. • The originalinformation is then demodulated back to its original form.
Signal Propagation reflection • Receiving power additionally influenced by • shadowing (e.g. through a wall or a door) • refraction depending on the density of a medium • reflection at large obstacles • scattering at small obstacles • diffraction at edges diffraction refraction scattering shadow fading 6
Multipath Fading • The signal takes many paths to the destination. The propagation delay along each path is different. • How many meters difference gives you 0.00001 seconds of delay difference? 2 3 1 The shorter path 0 2 3 1 0 The longer path Received signal – a combination of the two signals
Wireless Channel Characteristics • Radio propagation • Multipath, fade, attenuation, interference & capture • Received power is inversely proportional to the distance: distance-power gradient • Free space: factor 2 • In building corridors or large open indoor areas: <2 • Metal buildings: factor 6 • Recommended simulation factors: 2~3 for residential areas, offices and manufacturing floors; 4 for urban radio communications
Free-Space Isotropic(各向同性的) Signal Propagation In free space, receiving power proportional to 1/d²(d = distance between transmitter and receiver) Suppose transmitted signal is x,received signal y = h x, where h is proportional to 1/d² Reduction also depends on wavelength Long wave length (low frequency) has less loss Short wave length (high frequency) has more loss d Pr Pt • Pr: received power • Pt: transmitted power • Gr, Gt: receiver and transmitter antenna gain • (=c/f): wave length
Wireless Channel • Wireless transmission is error prone • Wireless error and contention are location dependent • Wireless channel capacity is also location dependent Channel
信道容量:香农定理 C：信道容量，信道可能传输的最大信息速率， 即信道所能达到的最大传输能力 B：信道带宽 S：信号平均功率 N：噪声平均功率 S/N: 信噪比
Link-Level Measurements • Measurements taken from 802.11b-based MIT Roofnet • Focus: • Explore reasons for loss • mainly on long outdoor links D. Aguayo, etc., "Link-level measurements from an 802.11b mesh network," ACM Sigcomm 2004.
Roofnet: multihop wireless mesh 1 kilometer
Using omni-direction antenna + Easy to deploy + Provide high connectivity - Don’t allow engineered link quality
Lossy radio links are common Broadcast packet delivery probability 70-100% 30-70% 1-30% 1 kilometer
Delivery prob. uniformly distributed Broadcast Packet Delivery Probability > two-thirds of links deliver less than 90% Node Pair
Implications • Protocols should exploit intermediate-quality links • 802.11 transmit bit-rate selection • Link-quality-aware routing (ETX, LQSR) • Opportunistic protocols (OMAC, ExOR) • An emerging research direction
Hypotheses for intermediate delivery probability • Marginal signal-to-noise ratios • Interference: Long bursts • Interference: Short bursts (802.11) • Multi-path interference
Methodology: Link-level measurements of packet loss • Goal: all-pairs loss rates • Each node broadcasts for 90 seconds • All other nodes listen • Raw link-level measurements: • No ACKs, retransmissions, RTS/CTS • No other Roofnet traffic • No 802.11 management frames • No carrier sense
Hypothesis 1: Marginal S/N • Simplified model for packet loss: • P(delivery) = f(signal/noise) • Signal strength reflects attenuation • Noise reflects interference • Perhaps marginal S/N explains intermediate delivery probabilities
Delivery vs. S/N with a cable and attenuator Broadcast packet delivery probability Laboratory Signal-to-noise ratio (dB)
Laboratory Roofnet Delivery vs. S/N on Roofnet Broadcast packet delivery probability Signal-to-noise ratio (dB) S/N does not predict delivery probability for intermediate-quality links
Hypothesis 2: long bursts of interference A B Bursty noise might corrupt packets without affecting S/N measurements
Loss over time on two different Roofnet links avg: 0.5 stddev: 0.28 Delivery probability avg: 0.5 stddev: 0.03 Time (seconds) The top graph is consistent with bursty interference. The bottom graph is not.
Hypothesis 3: short bursts of interference (802.11) A B • MAC doesn’t prevent all concurrent xmits • Outcome depends on relative signal levels • Hypothesis: When a nearby AP sends a packet, we lose a packet.
Methodology: record non-Roofnet 802.11 traffic • Goal: measure non-Roofnet traffic • Before the broadcast experiments • Each node records all 802.11 traffic
No correlation between foreign traffic observed and packets lost Experiment packets lost per second Non-Roofnet packets observed per second (before the experiment)
Hypothesis 4: Multi-path interference B B A Reflection is a delayed and attenuated copy of the signal
Receiver Sender A channel emulator to investigate multi-path effects delay attenuation
Reflection causes intermediate packet loss Delivery probability Delay of second ray (nanoseconds or feet)
Summary • Most Roofnet links have intermediate loss rates • SNR is related to packet loss rate, but does not predict delivery probability • Loss is not consistent withbursty interference • Multi-path is likely to be a major cause wireless environment is really different from wired, and is almost unpredictable.
Mobility • Why mobility? • 30~40% of the US workforce is mobile (Yankee group) • Hundreds of millions of users are already using portable computing devices and more than 60% of them are prepared to pay for wireless access to the backbone information
Mobility • Four types of activities for a typical office work during a workday: • Communication (fax, email) • Data manipulation (word processing, directory services, document access & retrieval) • Information access (database access and update, internet access and search) • Information share (groupware, shared file space) • Question: how does mobility affect each of the above activities?
Protocol Stack • Look at: • Applications/Services • OS issues • Middleware (skip): • Transcoding Application Layer Middleware and OS Transport Layer Network Layer Link Layer & Below
Issues in building services in mobile networking environments • Mobility induced issues: • Seamless services: service migration • Location services: location itself is a service • Heterogeneity induced issues: • Hardware diversity • Client devices & different networks • Software diversity • System software: OS, networking protocols • Application software • Wireless induced issues: • Time-varying network connectivity: disconnection, partial connection, full connection
Possible services for mobile environments • Location service • Location transparent services • Hide locations from users • Location dependent services • Services “local” to a geographic location • Not available globally • Location aware services • Services are globally available, but multiple instantiations of the same service are a function of locations • Service adapts to a location
Issues in Operating Systems • Energy-efficient scheduler • File systems for disconnected operation due to mobility and disconnected wireless links • access the same file as if connected • retain the same consistency semantics for shared files as if connected • availability and reliability as if connected • ACID (atomic/recoverability, consistent, isolated/ serializablity, durable) properties for transactions • Constraints: • disconnection and/or partial connection • low bandwidth connection • variable bandwidth and latency connection • connection cost
Next Step: Networking Issues
Physical/MAC Layer • Requirements: • Continuous access to the channel to transmit a frame without error • Fair access to the channel: how is fairness quantified? • Low power consumption • Increase channel throughput within the given frequency band • Constraints: • Channel is error prone • Channel contention and error are location dependent • Transmission range is limited (but also enables channel reuse) • Shared channel (hidden/exposed station problem)
Physical/MAC Layer • Possible options: • Physical layer: • Narrow band vs wide band: direct sequence, frequency hopping, OFDM • Antenna technology: smart antenna, directional antenna, MIMO • Adaptive modulation • MAC layer • Multiple access protocols (CSMA/CA, MACAW, etc.) • Frame reservation protocols (TDMA, DQRUMA, etc.)
Network Layer • Requirements: • Maintain connectivity while user roams • Allow IP to integrate transparently with roaming hosts • Address translation to map location-independent addressing to location dependent addressing • Packet forwarding • Location directory • Support multicast, anycast • Ability to switch interfaces on the fly to migrate between failure-prone networks • Ability to provide quality of service: what is QoS in this environment?
Network Layer • Constraint: • Unaware hosts running IP • Route management for mobile hosts needs to be dynamic • A backbone may not exist (ad-hoc network)
Network Layer • Possible options: • Mobile IP and its variants • Two-tier addressing (location independent addressing <-> location dependent addressing) • A smart forwarding agent which encapsulates packets from unware host to forward them to MH • Location directory for managing location updates • Ad hoc routing • Shortest path, source routing, multipath routing
Transport Layer • Requirements: • Congestion control and rate adaptation • Doing the right thing in the presence of different packet losses • Handling different losses (mobility-induced disconnection, channel, reroute) • Improve transient performance • Constraints: • Typically unware of mobility, yet affected by mobility • Packet may be lost due to congestion, channel error, handoffs, change of interfaces, rerouting failures • Link-layer and transport layer retransmit interactions
Transport Layer • Options: • Provide indirection • Make transport layer at the end hosts ware of mobility • Provide smarts in intermediate nodes (e.g. BS) to make lower-layer transport aware • Provide error-free link layers