230 likes | 503 Views
SOAR: Simple Opportunistic Adaptive Routing Protocol for Wireless Mesh Networks. Eric Rozner , Jayesh Seshadri , Yogita Ashok Mehta, and Lili Qiu Presented by Hamid Hamraz. Outlines. Brief Introduction to Wireless Mesh Networks Traditional and Opportunistic Routing SOAR Algorithm
E N D
SOAR: Simple Opportunistic Adaptive Routing Protocol for Wireless Mesh Networks Eric Rozner, JayeshSeshadri, Yogita Ashok Mehta, and LiliQiu Presented by Hamid Hamraz
Outlines • Brief Introduction to Wireless Mesh Networks • Traditional and Opportunistic Routing • SOAR Algorithm • Performance Analysis • conclusions
Wireless Mesh Networks (WMN) • WMNs an attractive easy-to-deploy communication paradigm: • Mesh nodes that are almost stationary and not constrained serve as a backbone • Other nodes get connected through mesh nodes (wired or wireless) • Routing protocol design is critical for performance and reliability: • Traditional routing • Opportunistic routing
Traditional Routing • Adopt similar techniques for wireline networks: • Select the best path for each source & destination pair according to some metrics • Hop count is the dominent metric • sending the data over the predetermined path • DSR, AODV, DSDV, LQSR
Opportunistic Routing • Hop count a suitable metric? • Wireless medium is unreliable and unpredictable • Not all hops are equal! • Other metrics taking the quality of link into account: • link loss rate • packet transmission time • signal-to-noise ratios
Opportunistic Routing • Exploits the broadcast nature of wireless communication • Does not commit to a particular path before data transmission • Route comparison based on link quality aware metrics • Sender broadcasts data and closest neighbor to destination is picked to forward the packet • Effectively combine multiple weak link to a strong one • Utilize the nature of the wireless links (transmissions may reach unexpectedly very far or unexpectedly not even very close)
Design Challenges • The goal is to maximize the packet progress in each transmission without duplicate transmissions or significant overhead • Candidate forwarding node selection: adjusting overhead • Avoid duplicate transmissions when multiple nodes overhear a transmission • Los Recovery: MAC does not handle losses for broadcast • Rate Control: avoiding the interference in the subsequent hops
SOAR Major Components • adaptive forwarding path selection to leverage path diversity while avoiding diverging paths; 2. priority timer-based forwarding to allow only the best forwarding node to forward the packet; 3. local loss recovery to efficiently detect and retransmit lost packets; 4. adaptive rate control to determine an appropriate sending rate according to the current network condition.
SOAR Overview • Every node priodically measures and disseminates the link state based on ETX • Based on this, the sender selects a default path and a list of eligible next-hop nodes • The sender broadcasts this data • Upon hearing, neighbors if not in the forwarding list: discard the packet • Otherwise: set a timer based on the proximity (ETX measured) to the destination to forward the packet thereafter– closer the node, smaller the timer • Any node hearing the packet transmission removes the packet from its queue to avoid duplicate transmissions
Default Path Selection • ETX measures the expected number of transmissions required to reliably transmit a packet across the link • Each node broadcasts probe packets regularly to measure and update the links’ ETXs with its neighbors • Links’ ETXs measured and disseminated through the network proactively • default path is the shortest path in terms of ETX
Forwarding List Selection • Select the forwarding list where: • The forwarding node (FN) is closer (smaller ETX) to the destination – ensures progress • The ETX to the FN is within a threshold – optimizing the probability of packet transmission to avoid retransmissions • Each FN is close to at least one node on the default path -with ETX below a threshold • The ETX between each pair of the FNs is within a threshold • The last two ensure FNs to have good connectivity between themselves and to the default path • FNs selection is done within each relay per each packet • Maximum number of FNs is bounded, e.g. to 5
More Optimization • Additional FNs not needed to be added if the closest ones to the destination are sufficiently reliable • Thus, start adding closest FN in an increasing order until: • Total certainty goes over a confidence level, or; • The maximum number of FNs reached • If the maximum reached and the level of certainty is below a level: • Replace the last FN with the closest node to the current node (not to the destination)
Priority-Based Forwarding • Neighbors hearing a packet drops it if not among the FNs • Based on their position in the ordered FN list, they set a timer to forward the packet; the farther in the order, the bigger the timer • Closer neighbors to the destination broadcast earlier • Others who are going to broadcast later, if hearing this, will simply drop the packet to avoid duplicate forwarding
Local Recovery • MAC does not provide certainty for broadcast • SOAR applies a per hop acknowlegement scheme – piggybacked AKS on data packets, and solo ACKs when no data transmission • Each FN tries to retransmit the packet until it receives ACK from a closer node or maximum number of retransmissions reached • Issues: • When to send ACK • What to include in the ACK • When to retransmit
Rate Control • Wireless interference degrades performance • Transport layer congestion control does not work well for multi-hop wireless networks • SOAR measures an end-to-end delay through an acknowledgement scheme in order to adjust the size of the transmission window • End-to-end ACKs are transferred via MAC unicast through the shortest path
Empirical Evaluations • Once NS-2 Simulater is used • A wireless framework of several PCs also utilized as the testbed to get more real-world results • A warm-up period is applied to let the network statistics, e.g. ETX converge • SOAR is compared with EXOR and shortest-path over:
Goodput Over Multiple Flows • Fig 18
conclusions • SOAR a novel opportunistic Routing which Judiciously selects the forwarding nodes • Employs priority based forwarding by setting timers • Detects packet los and tries to recover them • Employs rate control to avoid starvation due to wireless interference • Tries to avoid diverging path • Performs well when there is multiple simultaneous flows • The joint design of routing and rate limiting as in SOAR is useful, and may be useful to the design of other opportunistic routing protocols. • the default path selection algorithm is important to the performance of opportunistic routing. • ETX is utilized as the metric – any other metrics can be applied simply • Simulations and testbed experiments demonstrate the effectiveness of SOAR. • Enhancing default path selection algorithm as the future work