Accurate taxi demand-supply forecasting is a challenging application of ITS (Intelligent Transportation Systems), due to the complex spatial and temporal patterns involved. We investigate the impact of different spatial partitioning techniques on the prediction performance of an LSTM (Long Short-Term Memory) network, in the context of taxi demand-supply forecasting. We consider two tessellation schemes: (i) the variable-sized Voronoi tessellation, and (ii) the fixed-sized Geohash tessellation. While the widely employed ConvLSTM (Convolutional LSTM) method can model fixed-sized Geohash partitions, the standard convolutional filters cannot be applied on variable-sized Voronoi partitions. To explore the impact of the Voronoi strategy, we propose the use of a GraphLSTM (Graph-based LSTM) model, by representing the Voronoi spatial partitions as nodes on an arbitrarily structured graph. The GraphLSTM model offers competitive performance against the ConvLSTM model, at a lower computational complexity, across three real-world large-scale taxi demand-supply data sets, with different performance metrics. To ensure superior performance across diverse settings, a HEDGE based ensemble learning algorithm is applied over the ConvLSTM and the GraphLSTM networks.

A key problem in location-based modeling and forecasting lies in identifying suitable spatial and temporal resolutions. In particular, judicious spatial partitioning can play a significant role in enhancing the performance of location-based forecasting models. In this paper, we investigate two widely used tessellation strategies for partitioning city space, in the context of real-time taxi demand forecasting. Our study compares (1) the Geohash tessellation and (2) the Voronoi tessellation, using two distinct taxi demand data sets, over multiple time scales. For the purpose of comparison, we employ classical time-series tools to model the spatio-temporal demand. Our study finds that the performance of each tessellation strategy is highly dependent on the city geography, spatial distribution of the data, and the time of the day, and that neither strategy is found to perform optimally across the forecast horizon. We propose a combining algorithm that selects the best tessellation strategy at each time step, based on their recent performance. Our algorithm is a non-stationary variant of the well-known HEDGE algorithm for choosing the best advice from multiple experts. We show that the proposed strategy performs consistently better than either of the two tessellation strategies across the data sets considered, at multiple time scales, and with different performance metrics. We achieved an average accuracy of above 80% per km2for both data sets considered at 60 min aggregation levels.

In this paper, we analyze a model for transport control protocol (TCP) along with a non-adaptive virtual queue (VQ) and an adaptive virtual queue (AVQ) management policy. In the class of transport protocols, we focus on compound TCP as it is the default protocol in the Windows operating system. We start by conducting a local stability analysis for the underlying fluid models. For the VQ policy, we show that small virtual buffers play an important role in ensuring stability, whereas the AVQ policy could readily lose local stability as the link capacity, the feedback delay, or the link's damping factor gets large. With both the queue policies, the protocol parameters of compound TCP also influence stability. Furthermore, in both the models, we explicitly show that as parameters vary the loss of local stability would occur via a Hopf bifurcation. For the AVQ policy, we are also able to analytically verify if the Hopf bifurcation is super-critical, and determine the stability of the bifurcating limit cycles. Packet-level simulations, conducted over two topologies, using the network simulator (NS2) confirm the existence of stable limit cycles in the queue size.

We study Compound TCP (C-TCP), the default TCP in the Windows operating system, with Random Exponential Marking (REM) and the widely used Drop-Tail queue policy. The performance metrics we consider are stability of the queue size, queuing delay, link utilization, and packet loss. We analyze the following models: 1) a nonlinear model for C-TCP with Drop-Tail and small buffers; 2) a stochastic variant of REM along with C-TCP; and 3) the original REM proposal as a continuous-time nonlinear model with delayed feedback. We derive conditions to ensure local stability and show that variations in system parameters can induce a Hopf bifurcation, which would lead to the emergence of limit cycles. With Drop-Tail and small buffers, the Compound parameters and the buffer size both play a key role in ensuring stability. In the stochastic variant of REM, larger thresholds for marking/dropping packets can destabilize the system. With the original REM proposal, using Poincaré normal forms and the center manifold analysis, we also characterize the type of the Hopf bifurcation. This enables us to analytically verify the stability of the bifurcating limit cycles. Packet-level simulations corroborate some of the analysis. Some design guidelines to ensure stability and low latency are outlined.

Transport protocols that use explicit rate feedback are a promising alternative to the algorithms which use implicit, packet loss based feedback. A key issue in the design of rate controlled communication networks is how new flows may be admitted at a fair and high starting rate. In this paper, for the management of new flows we present a modified NetFPGA Rate Control Protocol (RCP) implementation for a recently proposed admission management process at the routers. Using this implementation, we perform an evaluation of the admission of new flows in rate controlled networks. We conduct experimental tests across a range of bandwidth-delay product environments, and also deal with the arrival of a large number of flows. Our work serves to exhibit that the proposed admission management process appears to be an attractive method for dealing with the admission of new flows while maintaining small queues. Additionally, the algorithm allows the routers to maintain a specified target link utilisation. A consequence of such small queues is that buffers, in routers, may be dimensioned to be much smaller than the current, bandwidth-delay product, buffer sizing rule.

In this article we work with the class of transport protocols that use queuing delay as a primary feedback signal for flow and congestion control. Our focus will be on this class of transport protocols operating over small buffered networks; i.e., buffers that are of the order of tens of packets. We show that the performance of Vegas and FAST can be very sensitive to the precise choice of small buffer size. With Reno and Vegas, or Reno and FAST, operating together we also exhibit that AIMD Reno dominates the dynamics of flow control. Some guidelines are suggested that could help design and enhance protocol performance in small buffered networks.