Efficient Multihop Broadcast for Wideband Systems Ivana Maric and Roy Yates Wireless Broadcast • Wireless network of N nodes • Source transmits with rate R • Messages are to be delivered to all the nodes • Nodes can choose a power level for each transmission • Problem: Broadcast at rate R to all nodes with minimum total power • Comment: N=3 nodes is a single relay channel • How many simplifications are needed? source System Model: Orthogonal Channels • A link: AWGN channel with bandwidth W • Large bandwidth resources •Each transmission in an orthogonal channel •Nodes can listen to all the channels • Motivation: Sensor networks • Low-powered nodes, very low data rates Minimum-energy broadcast problem • Min-energy broadcast tree problem [J. Wieselthier, G. Nguyen, A. Ephremides] • Wired network: Min-cost spanning tree problem • ‘Wireless multicast advantage’: all the nodes in the transmission range will benefit from a transmission • Problem is NP-complete [M. Čagalj et al., Ahluwalia et al., W. Liang] source source Accumulative broadcast • Allow nodes to collect energy of unreliably received signals Wireless advantage Accumulative broadcast Accumulative broadcast • As the message is forwarded, a node has multiple opportunities to receive energy needed for reliable reception of that message • Key issue: Who do you listen to? source Reliable forwarding • A node can forward a message only after reliable decoding • Disadvantage: suboptimal source • Benefits: •Simplifies the system architecture •Still allows for unreliable overheard information •Imposes an ordering on the node transmissions source Relays Use Repetition Coding • Relays resend the same codeword • A node m will decode a codeword using transmissions of a subset of nodes that became reliable prior to node m P1 X • After K nodes retransmit a codeword X: source P2 • Received signal for a symbol x: Y = hx + n X P3 Y X … PK m • Maximum rate: K I(x;y) = W log2(1 + Σ hmkPk/NoW) k=1 • X Upper bound:K CMAC = W Σ log2(1 + hmkPk/NoW) k=1 Repetition is OK for Large W • Given fixed powers {P1,…PK } and reliable forwarding, the maximum rate achievable from the source to any destination is achieved by the repetition coding in the limit of large W. • As W ∞, I(x;y) Σ hmkPk/Noln2 source P1 k P2 • MAC Upper bound: CMAC = W Σ log2(1 + hmkPk/NoW) Σ hmkPk/Noln2 k P3 m … PK • In such a network, how do we solve the broadcast problem? Difference from Min-Energy Broadcast Tree • The total transmit power of the minimum-energy broadcast tree upper bounds the total transmit power of accumulative broadcasting • • In the MBT problem, knowing the broadcast tree solves the problem completely: power levels are uniquely determined For accumulative broadcast, tree is not meaningful • Different total power for orders: 1-2-3 and 1-3-2 5 3 2 1 4 source source Approach • Divide the problem into two subproblems: • Choose a reliability schedule • • • • An order in which nodes become reliable Also, an order in which nodes are given a chance to transmit For each node, schedule specifies a subset of nodes that contribute to its reliable decoding Given a schedule, find the best power levels • Can be formulated as LP LP for Transmit Powers 5 4 2 1 3 source • Fix a schedule: 1 2 3 4 5 • We know which nodes contribute to the energy collected at a node min (p1 + p2 + p3 + p4) h21p1 ≥ PT h31p1 + h32p2 ≥ PT h41p1 + h42p2 + h43p3 ≥ PT h51p1 + h52p2 + h53p3 + h54p4 ≥ PT p1, p2, p3, p4 ≥0 • But, finding the optimal schedule is NP-complete. Greedy Filling Heuristic • • Given S={reliable nodes}, U={unreliable nodes} Choose node k to maximize “filling rate” of the unreliable nodes k = arg max Σ hji iєS jєU • Choose power Pk to make one more node reliable • Offline Optimization: • If node k transmits multiple times, Pk1, Pk2…, set Pk= Σ Pki and transmit once • Readjust schedule Simple Experiments • Throw N nodes in a square (100 trials) • Propagation exponent=2 • For small N: • Enumerate all schedules, find optimal AB powers • Compare with – greedy heuristic – BIP [Wieselthier, Nguyen, Ephremides] (ignores unreliable overheard messages) • For large N: • Compare greedy heuristic, BIP Performance results Note: Powers normalized by optimal sol’n Performance results Conclusion • Accumulative broadcast: Nodes collect energy of unreliably received signals • Only reliable forwarding is allowed • Large BW, relays can use repetition coding • Formulate problem as two subproblems: 1. Find a reliable schedule 2. LP to find the optimum power levels for a given schedule • Finding the optimum schedule is NP-complete • Proposed a heuristics to find a good schedule • Compared the algorithm performance with the optimum solution and BIP performance